diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36679 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 6109,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0869565217391305e-07,
+      "loss": 0.6885,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-07,
+      "loss": 0.8892,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2608695652173915e-07,
+      "loss": 0.7168,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 0.6086,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.434782608695653e-07,
+      "loss": 0.8313,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.521739130434783e-07,
+      "loss": 0.7678,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.608695652173914e-07,
+      "loss": 0.7828,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 0.6011,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.782608695652175e-07,
+      "loss": 0.7817,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0869565217391306e-06,
+      "loss": 0.698,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1956521739130436e-06,
+      "loss": 0.7114,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 0.6857,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4130434782608697e-06,
+      "loss": 0.7539,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.521739130434783e-06,
+      "loss": 0.7083,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6304347826086957e-06,
+      "loss": 0.6846,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 0.6377,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8478260869565218e-06,
+      "loss": 0.6704,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.956521739130435e-06,
+      "loss": 0.7356,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.065217391304348e-06,
+      "loss": 0.6562,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 0.7046,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.282608695652174e-06,
+      "loss": 0.6667,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.391304347826087e-06,
+      "loss": 0.7588,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7375,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 0.7856,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7173913043478263e-06,
+      "loss": 0.7578,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8260869565217393e-06,
+      "loss": 0.6606,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9347826086956528e-06,
+      "loss": 0.7039,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.043478260869566e-06,
+      "loss": 0.6816,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.152173913043479e-06,
+      "loss": 0.6597,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2608695652173914e-06,
+      "loss": 0.567,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3695652173913045e-06,
+      "loss": 0.8062,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 0.77,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5869565217391305e-06,
+      "loss": 0.595,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6956521739130436e-06,
+      "loss": 0.6123,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.804347826086957e-06,
+      "loss": 0.7615,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91304347826087e-06,
+      "loss": 0.6909,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.021739130434783e-06,
+      "loss": 0.7686,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.130434782608696e-06,
+      "loss": 0.7131,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.239130434782609e-06,
+      "loss": 0.748,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.5852,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.456521739130435e-06,
+      "loss": 0.6538,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.565217391304348e-06,
+      "loss": 0.7388,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.673913043478261e-06,
+      "loss": 0.7,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782608695652174e-06,
+      "loss": 0.7485,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.891304347826087e-06,
+      "loss": 0.6206,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7615,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.108695652173914e-06,
+      "loss": 0.7012,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 0.7031,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3260869565217395e-06,
+      "loss": 0.7935,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4347826086956525e-06,
+      "loss": 0.6633,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.543478260869566e-06,
+      "loss": 0.6802,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 0.7297,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.760869565217392e-06,
+      "loss": 0.7122,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8695652173913055e-06,
+      "loss": 0.6321,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.978260869565218e-06,
+      "loss": 0.7603,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.8186,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.195652173913044e-06,
+      "loss": 0.7727,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.304347826086958e-06,
+      "loss": 0.6785,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.41304347826087e-06,
+      "loss": 0.8384,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 0.7153,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.630434782608696e-06,
+      "loss": 0.7578,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.739130434782609e-06,
+      "loss": 0.7891,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.847826086956523e-06,
+      "loss": 0.6655,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.7656,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.065217391304349e-06,
+      "loss": 0.7065,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.173913043478261e-06,
+      "loss": 0.7129,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.282608695652175e-06,
+      "loss": 0.8296,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 0.9595,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.6182,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.608695652173914e-06,
+      "loss": 0.7258,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.717391304347827e-06,
+      "loss": 0.7354,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.585,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.934782608695653e-06,
+      "loss": 0.6829,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.043478260869566e-06,
+      "loss": 0.7764,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.15217391304348e-06,
+      "loss": 0.7524,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 0.6733,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.369565217391305e-06,
+      "loss": 0.6082,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.478260869565218e-06,
+      "loss": 0.73,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.586956521739131e-06,
+      "loss": 0.7896,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.7805,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.804347826086957e-06,
+      "loss": 0.8718,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91304347826087e-06,
+      "loss": 0.804,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.021739130434784e-06,
+      "loss": 0.7705,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 0.7603,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.23913043478261e-06,
+      "loss": 0.6763,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.347826086956523e-06,
+      "loss": 0.6521,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.456521739130436e-06,
+      "loss": 0.6726,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.8203,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.673913043478262e-06,
+      "loss": 0.533,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.782608695652175e-06,
+      "loss": 0.6821,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.891304347826088e-06,
+      "loss": 0.8647,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.6282,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0108695652173914e-05,
+      "loss": 0.6499,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0217391304347829e-05,
+      "loss": 0.7114,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.032608695652174e-05,
+      "loss": 0.6906,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 0.7666,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0543478260869566e-05,
+      "loss": 0.5901,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0652173913043479e-05,
+      "loss": 0.7522,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076086956521739e-05,
+      "loss": 0.6658,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.6724,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0978260869565218e-05,
+      "loss": 0.6365,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1086956521739131e-05,
+      "loss": 0.5735,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1195652173913046e-05,
+      "loss": 0.6118,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.6697,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.141304347826087e-05,
+      "loss": 0.7622,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1521739130434783e-05,
+      "loss": 0.6951,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1630434782608698e-05,
+      "loss": 0.5515,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1739130434782611e-05,
+      "loss": 0.7329,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1847826086956522e-05,
+      "loss": 0.6992,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1956521739130435e-05,
+      "loss": 0.5842,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2065217391304348e-05,
+      "loss": 0.6902,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 0.5648,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2282608695652175e-05,
+      "loss": 0.7808,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2391304347826088e-05,
+      "loss": 0.6274,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.79,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2608695652173915e-05,
+      "loss": 0.5559,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2717391304347828e-05,
+      "loss": 0.7075,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.282608695652174e-05,
+      "loss": 0.595,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2934782608695653e-05,
+      "loss": 0.7322,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.6646,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.315217391304348e-05,
+      "loss": 0.6067,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3260869565217392e-05,
+      "loss": 0.7419,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3369565217391305e-05,
+      "loss": 0.791,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 0.8218,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3586956521739133e-05,
+      "loss": 0.6665,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3695652173913046e-05,
+      "loss": 0.8789,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3804347826086957e-05,
+      "loss": 0.7632,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.6516,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4021739130434783e-05,
+      "loss": 0.6882,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4130434782608698e-05,
+      "loss": 0.7793,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.423913043478261e-05,
+      "loss": 0.8086,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 0.7979,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4456521739130435e-05,
+      "loss": 0.7405,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.456521739130435e-05,
+      "loss": 0.6616,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4673913043478263e-05,
+      "loss": 0.7734,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 0.771,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4891304347826087e-05,
+      "loss": 0.8159,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6272,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5108695652173915e-05,
+      "loss": 0.7363,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 0.7373,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.532608695652174e-05,
+      "loss": 0.606,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5434782608695654e-05,
+      "loss": 0.7354,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5543478260869567e-05,
+      "loss": 0.6641,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.5994,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576086956521739e-05,
+      "loss": 0.6271,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5869565217391306e-05,
+      "loss": 0.6804,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.597826086956522e-05,
+      "loss": 0.8462,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 0.8452,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6195652173913045e-05,
+      "loss": 0.7717,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.630434782608696e-05,
+      "loss": 0.6194,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.641304347826087e-05,
+      "loss": 0.606,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 0.8438,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6630434782608698e-05,
+      "loss": 0.6345,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673913043478261e-05,
+      "loss": 0.8438,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6847826086956524e-05,
+      "loss": 0.738,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6956521739130437e-05,
+      "loss": 0.752,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.706521739130435e-05,
+      "loss": 0.6567,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7173913043478263e-05,
+      "loss": 0.6118,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7282608695652176e-05,
+      "loss": 0.7383,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.8611,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.7341,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7608695652173915e-05,
+      "loss": 0.6941,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7717391304347828e-05,
+      "loss": 0.6378,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.782608695652174e-05,
+      "loss": 0.6936,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7934782608695654e-05,
+      "loss": 0.689,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8043478260869567e-05,
+      "loss": 0.603,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.815217391304348e-05,
+      "loss": 0.6021,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.6675,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8369565217391306e-05,
+      "loss": 0.6221,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.847826086956522e-05,
+      "loss": 0.71,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8586956521739132e-05,
+      "loss": 0.7612,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 0.6621,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8804347826086958e-05,
+      "loss": 0.7756,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891304347826087e-05,
+      "loss": 0.6169,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9021739130434784e-05,
+      "loss": 0.7571,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.8633,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.923913043478261e-05,
+      "loss": 0.6238,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9347826086956523e-05,
+      "loss": 0.6958,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9456521739130436e-05,
+      "loss": 0.7014,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 0.7197,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9673913043478263e-05,
+      "loss": 0.6711,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9782608695652176e-05,
+      "loss": 0.7412,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.989130434782609e-05,
+      "loss": 0.7607,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.6665,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998594298752e-05,
+      "loss": 0.5906,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994377195405e-05,
+      "loss": 0.6179,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999987348691148e-05,
+      "loss": 0.7673,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997750878795e-05,
+      "loss": 0.7734,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996485748858e-05,
+      "loss": 0.6475,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999949394796597e-05,
+      "loss": 0.6587,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999931120716343e-05,
+      "loss": 0.7573,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999910035252965e-05,
+      "loss": 0.6562,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988613841238e-05,
+      "loss": 0.6621,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999985943020131e-05,
+      "loss": 0.5258,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999982991062727e-05,
+      "loss": 0.7043,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979757969855e-05,
+      "loss": 0.6692,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999762437424247e-05,
+      "loss": 0.6475,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972448381424e-05,
+      "loss": 0.5703,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999683718879195e-05,
+      "loss": 0.709,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999640142630574e-05,
+      "loss": 0.8003,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999959375508063e-05,
+      "loss": 0.7578,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999544556242404e-05,
+      "loss": 0.6719,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999492546129723e-05,
+      "loss": 0.7466,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999437724757217e-05,
+      "loss": 0.5955,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999380092140294e-05,
+      "loss": 0.7749,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999319648295157e-05,
+      "loss": 0.6226,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992563932388e-05,
+      "loss": 0.6226,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999190326989005e-05,
+      "loss": 0.6963,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999121449564347e-05,
+      "loss": 0.6082,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999904976098419e-05,
+      "loss": 0.613,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998975261268693e-05,
+      "loss": 0.7202,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998897950438788e-05,
+      "loss": 0.6367,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998817828516226e-05,
+      "loss": 0.6958,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998734895523525e-05,
+      "loss": 0.8232,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998649151484e-05,
+      "loss": 0.6968,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999856059642176e-05,
+      "loss": 0.6975,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998469230361696e-05,
+      "loss": 0.7007,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998375053329502e-05,
+      "loss": 0.7407,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999827806535165e-05,
+      "loss": 0.55,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998178266455407e-05,
+      "loss": 0.7207,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998075656668834e-05,
+      "loss": 0.729,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999797023602078e-05,
+      "loss": 0.8694,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997862004540872e-05,
+      "loss": 0.6646,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999775096225955e-05,
+      "loss": 0.6482,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999763710920803e-05,
+      "loss": 0.5322,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997520445418317e-05,
+      "loss": 0.6394,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997400970923214e-05,
+      "loss": 0.8291,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997278685756304e-05,
+      "loss": 0.6616,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997153589951973e-05,
+      "loss": 0.731,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999702568354539e-05,
+      "loss": 0.7495,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999689496657251e-05,
+      "loss": 0.7073,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996761439070087e-05,
+      "loss": 0.7886,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996625101075658e-05,
+      "loss": 0.7021,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996485952627554e-05,
+      "loss": 0.5498,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996343993764896e-05,
+      "loss": 0.7939,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999619922452759e-05,
+      "loss": 0.7341,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999605164495635e-05,
+      "loss": 0.6523,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995901255092648e-05,
+      "loss": 0.7329,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995748054978778e-05,
+      "loss": 0.7354,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995592044657807e-05,
+      "loss": 0.8579,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995433224173594e-05,
+      "loss": 0.7183,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995271593570788e-05,
+      "loss": 0.7349,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995107152894834e-05,
+      "loss": 0.6584,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994939902191964e-05,
+      "loss": 0.7432,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994769841509197e-05,
+      "loss": 0.8352,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994596970894337e-05,
+      "loss": 0.7561,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994421290395997e-05,
+      "loss": 0.686,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994242800063565e-05,
+      "loss": 0.5674,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994061499947213e-05,
+      "loss": 0.7007,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999387739009792e-05,
+      "loss": 0.9185,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993690470567447e-05,
+      "loss": 0.7246,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999350074140834e-05,
+      "loss": 0.688,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993308202673936e-05,
+      "loss": 0.687,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999311285441838e-05,
+      "loss": 0.7847,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999291469669658e-05,
+      "loss": 0.6426,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992713729564248e-05,
+      "loss": 0.614,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992509953077888e-05,
+      "loss": 0.7461,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992303367294784e-05,
+      "loss": 0.6562,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999209397227302e-05,
+      "loss": 0.8047,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991881768071464e-05,
+      "loss": 0.6013,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991666754749776e-05,
+      "loss": 0.7422,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991448932368404e-05,
+      "loss": 0.7253,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.8579,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999100486067235e-05,
+      "loss": 0.6731,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999077861148252e-05,
+      "loss": 0.667,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990549553482694e-05,
+      "loss": 0.7339,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990317686737272e-05,
+      "loss": 0.7744,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999008301131145e-05,
+      "loss": 0.7954,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989845527271195e-05,
+      "loss": 0.6484,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989605234683275e-05,
+      "loss": 0.6201,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989362133615248e-05,
+      "loss": 0.5686,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998911622413546e-05,
+      "loss": 0.6267,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988867506313047e-05,
+      "loss": 0.6218,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988615980217926e-05,
+      "loss": 0.7708,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998836164592082e-05,
+      "loss": 0.6096,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998810450349323e-05,
+      "loss": 0.6895,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987844553007444e-05,
+      "loss": 0.7183,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987581794536552e-05,
+      "loss": 0.6936,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987316228154424e-05,
+      "loss": 0.8164,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998704785393572e-05,
+      "loss": 0.6335,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998677667195589e-05,
+      "loss": 0.7129,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986502682291173e-05,
+      "loss": 0.5745,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986225885018604e-05,
+      "loss": 0.729,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985946280215996e-05,
+      "loss": 0.4524,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985663867961962e-05,
+      "loss": 0.7959,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985378648335897e-05,
+      "loss": 0.7578,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985090621417985e-05,
+      "loss": 0.7214,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998479978728921e-05,
+      "loss": 0.6292,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984506146031327e-05,
+      "loss": 0.7305,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984209697726896e-05,
+      "loss": 0.6685,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983910442459262e-05,
+      "loss": 0.6526,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983608380312557e-05,
+      "loss": 0.7922,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99833035113717e-05,
+      "loss": 0.7468,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99829958357224e-05,
+      "loss": 0.6472,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982685353451162e-05,
+      "loss": 0.655,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982372064645274e-05,
+      "loss": 0.6455,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982055969392816e-05,
+      "loss": 0.6909,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998173706778265e-05,
+      "loss": 0.665,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981415359904437e-05,
+      "loss": 0.6398,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998109084584862e-05,
+      "loss": 0.6157,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980763525706427e-05,
+      "loss": 0.6631,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980433399569893e-05,
+      "loss": 0.6218,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998010046753182e-05,
+      "loss": 0.6409,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979764729685812e-05,
+      "loss": 0.5754,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979426186126263e-05,
+      "loss": 0.675,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979084836948342e-05,
+      "loss": 0.5767,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978740682248022e-05,
+      "loss": 0.7437,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978393722122055e-05,
+      "loss": 0.6707,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997804395666799e-05,
+      "loss": 0.7236,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977691385984158e-05,
+      "loss": 0.6831,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977336010169682e-05,
+      "loss": 0.6799,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976977829324466e-05,
+      "loss": 0.6638,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976616843549218e-05,
+      "loss": 0.8201,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976253052945426e-05,
+      "loss": 0.6924,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975886457615355e-05,
+      "loss": 0.6519,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975517057662078e-05,
+      "loss": 0.6228,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997514485318945e-05,
+      "loss": 0.7292,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974769844302105e-05,
+      "loss": 0.7368,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974392031105484e-05,
+      "loss": 0.7397,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974011413705795e-05,
+      "loss": 0.5588,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997362799221005e-05,
+      "loss": 0.6401,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997324176672604e-05,
+      "loss": 0.6274,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997285273736236e-05,
+      "loss": 0.8271,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972460904228365e-05,
+      "loss": 0.7676,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997206626743423e-05,
+      "loss": 0.5781,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971668827090895e-05,
+      "loss": 0.6204,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99712685833101e-05,
+      "loss": 0.6411,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997086553620436e-05,
+      "loss": 0.7534,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970459685887002e-05,
+      "loss": 0.4364,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997005103247212e-05,
+      "loss": 0.7456,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969639576074608e-05,
+      "loss": 0.615,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969225316810133e-05,
+      "loss": 0.751,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968808254795164e-05,
+      "loss": 0.7051,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996838839014696e-05,
+      "loss": 0.6665,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967965722983556e-05,
+      "loss": 0.8027,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996754025342378e-05,
+      "loss": 0.6499,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967111981587256e-05,
+      "loss": 0.5808,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996668090759438e-05,
+      "loss": 0.6838,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966247031566347e-05,
+      "loss": 0.6941,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965810353625134e-05,
+      "loss": 0.6997,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965370873893516e-05,
+      "loss": 0.7178,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.6821,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964483509554064e-05,
+      "loss": 0.6741,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964035625195702e-05,
+      "loss": 0.7532,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963584939545878e-05,
+      "loss": 0.8171,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99631314527313e-05,
+      "loss": 0.6631,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996267516487946e-05,
+      "loss": 0.6021,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996221607611864e-05,
+      "loss": 0.7378,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961754186577906e-05,
+      "loss": 0.6665,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996128949638711e-05,
+      "loss": 0.6213,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960822005676907e-05,
+      "loss": 0.6895,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960351714578718e-05,
+      "loss": 0.7466,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959878623224764e-05,
+      "loss": 0.7544,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995940273174805e-05,
+      "loss": 0.7678,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958924040282365e-05,
+      "loss": 0.7368,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995844254896229e-05,
+      "loss": 0.7656,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99579582579232e-05,
+      "loss": 0.6895,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957471167301232e-05,
+      "loss": 0.6501,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956981277233342e-05,
+      "loss": 0.7258,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995648858785725e-05,
+      "loss": 0.6521,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995599309931147e-05,
+      "loss": 0.6584,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955494811735308e-05,
+      "loss": 0.6877,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954993725268856e-05,
+      "loss": 0.7769,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995448984005298e-05,
+      "loss": 0.6069,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953983156229348e-05,
+      "loss": 0.709,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995347367394041e-05,
+      "loss": 0.7317,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99529613933294e-05,
+      "loss": 0.583,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952446314540337e-05,
+      "loss": 0.7275,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995192843771804e-05,
+      "loss": 0.6169,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951407763008098e-05,
+      "loss": 0.6714,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995088429055689e-05,
+      "loss": 0.7009,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950358020511597e-05,
+      "loss": 0.6277,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949828953020164e-05,
+      "loss": 0.7397,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949297088231337e-05,
+      "loss": 0.8511,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948762426294647e-05,
+      "loss": 0.8188,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948224967360407e-05,
+      "loss": 0.7236,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947684711579716e-05,
+      "loss": 0.7576,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947141659104464e-05,
+      "loss": 0.7034,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946595810087324e-05,
+      "loss": 0.6782,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994604716468176e-05,
+      "loss": 0.7173,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994549572304201e-05,
+      "loss": 0.6824,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944941485323114e-05,
+      "loss": 0.7998,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944384451680886e-05,
+      "loss": 0.7083,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943824622271934e-05,
+      "loss": 0.6184,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994326199725365e-05,
+      "loss": 0.7095,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942696576784203e-05,
+      "loss": 0.7627,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942128361022566e-05,
+      "loss": 0.6572,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941557350128477e-05,
+      "loss": 0.583,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940983544262475e-05,
+      "loss": 0.725,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994040694358588e-05,
+      "loss": 0.5825,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939827548260793e-05,
+      "loss": 0.6199,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939245358450117e-05,
+      "loss": 0.7322,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938660374317517e-05,
+      "loss": 0.6167,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938072596027464e-05,
+      "loss": 0.7759,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937482023745197e-05,
+      "loss": 0.7146,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936888657636756e-05,
+      "loss": 0.7229,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993629249786896e-05,
+      "loss": 0.8262,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935693544609415e-05,
+      "loss": 0.6018,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935091798026508e-05,
+      "loss": 0.7124,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993448725828941e-05,
+      "loss": 0.6174,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993387992556809e-05,
+      "loss": 0.7168,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933269800033287e-05,
+      "loss": 0.5345,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932656881856534e-05,
+      "loss": 0.6606,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932041171210153e-05,
+      "loss": 0.6897,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931422668267234e-05,
+      "loss": 0.6326,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930801373201668e-05,
+      "loss": 0.6804,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993017728618813e-05,
+      "loss": 0.7144,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992955040740207e-05,
+      "loss": 0.6714,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928920737019735e-05,
+      "loss": 0.6765,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928288275218146e-05,
+      "loss": 0.6831,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927653022175116e-05,
+      "loss": 0.6948,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927014978069235e-05,
+      "loss": 0.5652,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992637414307989e-05,
+      "loss": 0.7307,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992573051738724e-05,
+      "loss": 0.7622,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925084101172234e-05,
+      "loss": 0.6545,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992443489461661e-05,
+      "loss": 0.6467,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923782897902884e-05,
+      "loss": 0.8643,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992312811121436e-05,
+      "loss": 0.7192,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922470534735124e-05,
+      "loss": 0.6658,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921810168650044e-05,
+      "loss": 0.6782,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.7366,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992048106840577e-05,
+      "loss": 0.5789,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919812334620235e-05,
+      "loss": 0.6957,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991914081197619e-05,
+      "loss": 0.5886,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918466500662418e-05,
+      "loss": 0.7075,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917789400868504e-05,
+      "loss": 0.7031,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99171095127848e-05,
+      "loss": 0.8008,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916426836602457e-05,
+      "loss": 0.748,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915741372513396e-05,
+      "loss": 0.7883,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915053120710333e-05,
+      "loss": 0.6968,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914362081386764e-05,
+      "loss": 0.8154,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913668254736968e-05,
+      "loss": 0.6716,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912971640956002e-05,
+      "loss": 0.675,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912272240239715e-05,
+      "loss": 0.6543,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991157005278474e-05,
+      "loss": 0.656,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910865078788488e-05,
+      "loss": 0.6033,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910157318449155e-05,
+      "loss": 0.5505,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990944677196572e-05,
+      "loss": 0.6099,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990873343953795e-05,
+      "loss": 0.675,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908017321366387e-05,
+      "loss": 0.7629,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907298417652366e-05,
+      "loss": 0.6816,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906576728597993e-05,
+      "loss": 0.7014,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905852254406167e-05,
+      "loss": 0.6807,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990512499528057e-05,
+      "loss": 0.5898,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990439495142566e-05,
+      "loss": 0.7473,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990366212304668e-05,
+      "loss": 0.635,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902926510349663e-05,
+      "loss": 0.6235,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902188113541417e-05,
+      "loss": 0.5432,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901446932829533e-05,
+      "loss": 0.7129,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900702968422387e-05,
+      "loss": 0.5464,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899956220529143e-05,
+      "loss": 0.718,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899206689359733e-05,
+      "loss": 0.6753,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989845437512489e-05,
+      "loss": 0.6436,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989769927803611e-05,
+      "loss": 0.5991,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896941398305688e-05,
+      "loss": 0.5806,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896180736146688e-05,
+      "loss": 0.7073,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895417291772973e-05,
+      "loss": 0.6765,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989465106539917e-05,
+      "loss": 0.7368,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893882057240698e-05,
+      "loss": 0.655,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893110267513755e-05,
+      "loss": 0.7136,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892335696435326e-05,
+      "loss": 0.6467,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989155834422317e-05,
+      "loss": 0.6636,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890778211095836e-05,
+      "loss": 0.6914,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889995297272648e-05,
+      "loss": 0.6001,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988920960297371e-05,
+      "loss": 0.647,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988842112841992e-05,
+      "loss": 0.8076,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887629873832946e-05,
+      "loss": 0.6528,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886835839435248e-05,
+      "loss": 0.6147,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886039025450053e-05,
+      "loss": 0.6494,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988523943210138e-05,
+      "loss": 0.7354,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884437059614026e-05,
+      "loss": 0.5549,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883631908213573e-05,
+      "loss": 0.7544,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988282397812638e-05,
+      "loss": 0.7134,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882013269579585e-05,
+      "loss": 0.583,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881199782801117e-05,
+      "loss": 0.7148,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880383518019676e-05,
+      "loss": 0.645,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987956447546475e-05,
+      "loss": 0.668,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98787426553666e-05,
+      "loss": 0.7393,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987791805795628e-05,
+      "loss": 0.6969,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987709068346561e-05,
+      "loss": 0.5669,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98762605321272e-05,
+      "loss": 0.5437,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875427604174442e-05,
+      "loss": 0.6577,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874591899841504e-05,
+      "loss": 0.6616,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873753419363336e-05,
+      "loss": 0.6731,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872912162975668e-05,
+      "loss": 0.6401,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987206813091501e-05,
+      "loss": 0.6228,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871221323418658e-05,
+      "loss": 0.6401,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987037174072468e-05,
+      "loss": 0.697,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986951938307193e-05,
+      "loss": 0.6887,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868664250700034e-05,
+      "loss": 0.6034,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867806343849413e-05,
+      "loss": 0.7407,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866945662761252e-05,
+      "loss": 0.6589,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986608220767753e-05,
+      "loss": 0.9077,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986521597884099e-05,
+      "loss": 0.8687,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864346976495172e-05,
+      "loss": 0.6069,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863475200884386e-05,
+      "loss": 0.7952,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862600652253715e-05,
+      "loss": 0.6736,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861723330849043e-05,
+      "loss": 0.5754,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986084323691701e-05,
+      "loss": 0.6914,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.5581,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859074732461373e-05,
+      "loss": 0.5493,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858186322434964e-05,
+      "loss": 0.6909,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857295140875596e-05,
+      "loss": 0.6885,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985640118803381e-05,
+      "loss": 0.551,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985550446416094e-05,
+      "loss": 0.5278,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854604969509082e-05,
+      "loss": 0.656,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853702704331126e-05,
+      "loss": 0.6362,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852797668880735e-05,
+      "loss": 0.6643,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851889863412347e-05,
+      "loss": 0.5662,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850979288181188e-05,
+      "loss": 0.696,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850065943443253e-05,
+      "loss": 0.8154,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984914982945532e-05,
+      "loss": 0.5518,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984823094647495e-05,
+      "loss": 0.6653,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847309294760473e-05,
+      "loss": 0.6492,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846384874571005e-05,
+      "loss": 0.7336,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845457686166436e-05,
+      "loss": 0.7268,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844527729807436e-05,
+      "loss": 0.7585,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843595005755457e-05,
+      "loss": 0.7024,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842659514272724e-05,
+      "loss": 0.6807,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841721255622235e-05,
+      "loss": 0.5862,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984078023006778e-05,
+      "loss": 0.6616,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839836437873914e-05,
+      "loss": 0.7021,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983888987930598e-05,
+      "loss": 0.751,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983794055463009e-05,
+      "loss": 0.7417,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983698846411314e-05,
+      "loss": 0.6528,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98360336080228e-05,
+      "loss": 0.6416,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983507598662751e-05,
+      "loss": 0.7188,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983411560019651e-05,
+      "loss": 0.5962,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833152448999796e-05,
+      "loss": 0.5872,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983218653330815e-05,
+      "loss": 0.5581,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983121785339313e-05,
+      "loss": 0.6421,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983024640952707e-05,
+      "loss": 0.6089,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829272201983077e-05,
+      "loss": 0.7217,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828295231035054e-05,
+      "loss": 0.7354,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827315496957652e-05,
+      "loss": 0.6963,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982633300002632e-05,
+      "loss": 0.6465,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982534774051728e-05,
+      "loss": 0.7217,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982435971870752e-05,
+      "loss": 0.6897,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823368934874824e-05,
+      "loss": 0.7065,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822375389297735e-05,
+      "loss": 0.6812,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821379082255575e-05,
+      "loss": 0.5674,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982038001402845e-05,
+      "loss": 0.7205,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981937818489724e-05,
+      "loss": 0.772,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98183735951436e-05,
+      "loss": 0.7432,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981736624504995e-05,
+      "loss": 0.5347,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816356134899513e-05,
+      "loss": 0.6379,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981534326497626e-05,
+      "loss": 0.8779,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814327635564953e-05,
+      "loss": 0.7527,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813309246951124e-05,
+      "loss": 0.6101,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981228809942109e-05,
+      "loss": 0.6306,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981126419326193e-05,
+      "loss": 0.6489,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810237528761507e-05,
+      "loss": 0.5913,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980920810620846e-05,
+      "loss": 0.7429,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808175925892195e-05,
+      "loss": 0.6213,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807140988102905e-05,
+      "loss": 0.7898,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806103293131552e-05,
+      "loss": 0.7178,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805062841269875e-05,
+      "loss": 0.6094,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980401963281038e-05,
+      "loss": 0.6304,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802973668046364e-05,
+      "loss": 0.6533,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801924947271883e-05,
+      "loss": 0.6221,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800873470781778e-05,
+      "loss": 0.5864,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979981923887166e-05,
+      "loss": 0.6924,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979876225183792e-05,
+      "loss": 0.6633,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979770250997771e-05,
+      "loss": 0.7748,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796640013588976e-05,
+      "loss": 0.6431,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795574762970426e-05,
+      "loss": 0.7012,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979450675842154e-05,
+      "loss": 0.5999,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979343600024258e-05,
+      "loss": 0.8154,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792362488734585e-05,
+      "loss": 0.6519,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791286224199355e-05,
+      "loss": 0.7429,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979020720693947e-05,
+      "loss": 0.77,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978912543725829e-05,
+      "loss": 0.634,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978804091545995e-05,
+      "loss": 0.6829,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786953641849336e-05,
+      "loss": 0.6719,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785863616732138e-05,
+      "loss": 0.6274,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97847708404148e-05,
+      "loss": 0.7427,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783675313204547e-05,
+      "loss": 0.7415,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978257703540938e-05,
+      "loss": 0.6294,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.717,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780372229300135e-05,
+      "loss": 0.6355,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779265701605925e-05,
+      "loss": 0.647,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778156424566513e-05,
+      "loss": 0.614,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777044398493766e-05,
+      "loss": 0.7446,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775929623700318e-05,
+      "loss": 0.5845,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774812100499577e-05,
+      "loss": 0.8003,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977369182920572e-05,
+      "loss": 0.6372,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977256881013371e-05,
+      "loss": 0.7593,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771443043599262e-05,
+      "loss": 0.6689,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770314529918883e-05,
+      "loss": 0.7009,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976918326940984e-05,
+      "loss": 0.5341,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768049262390172e-05,
+      "loss": 0.6965,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97669125091787e-05,
+      "loss": 0.6909,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976577301009501e-05,
+      "loss": 0.6729,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764630765459455e-05,
+      "loss": 0.5642,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763485775593176e-05,
+      "loss": 0.7847,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762338040818067e-05,
+      "loss": 0.8545,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976118756145681e-05,
+      "loss": 0.624,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760034337832844e-05,
+      "loss": 0.613,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758878370270388e-05,
+      "loss": 0.5273,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757719659094437e-05,
+      "loss": 0.6284,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975655820463074e-05,
+      "loss": 0.781,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755394007205842e-05,
+      "loss": 0.6724,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754227067147038e-05,
+      "loss": 0.6456,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97530573847824e-05,
+      "loss": 0.5437,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975188496044078e-05,
+      "loss": 0.7773,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750709794451786e-05,
+      "loss": 0.6204,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974953188714581e-05,
+      "loss": 0.689,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748351238854006e-05,
+      "loss": 0.752,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747167849908305e-05,
+      "loss": 0.6565,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97459817206414e-05,
+      "loss": 0.6729,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744792851386764e-05,
+      "loss": 0.9136,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743601242478636e-05,
+      "loss": 0.6511,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742406894252024e-05,
+      "loss": 0.5845,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741209807042707e-05,
+      "loss": 0.6628,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740009981187237e-05,
+      "loss": 0.6426,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738807417022927e-05,
+      "loss": 0.6257,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973760211488787e-05,
+      "loss": 0.8569,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973639407512093e-05,
+      "loss": 0.6895,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735183298061726e-05,
+      "loss": 0.6675,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733969784050667e-05,
+      "loss": 0.6372,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973275353342891e-05,
+      "loss": 0.7581,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97315345465384e-05,
+      "loss": 0.6628,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730312823721834e-05,
+      "loss": 0.6975,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97290883653227e-05,
+      "loss": 0.6992,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972786117168523e-05,
+      "loss": 0.6208,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972663124315445e-05,
+      "loss": 0.6509,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725398580076132e-05,
+      "loss": 0.5483,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972416318279683e-05,
+      "loss": 0.7297,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972292505166387e-05,
+      "loss": 0.647,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721684187025333e-05,
+      "loss": 0.7812,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720440589230075e-05,
+      "loss": 0.5703,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971919425862773e-05,
+      "loss": 0.6392,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717945195568686e-05,
+      "loss": 0.7388,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97166934004041e-05,
+      "loss": 0.657,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715438873485913e-05,
+      "loss": 0.5154,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714181615166815e-05,
+      "loss": 0.7322,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712921625800277e-05,
+      "loss": 0.6172,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711658905740528e-05,
+      "loss": 0.6162,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710393455342572e-05,
+      "loss": 0.6926,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709125274962178e-05,
+      "loss": 0.6443,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707854364955884e-05,
+      "loss": 0.7112,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706580725680987e-05,
+      "loss": 0.856,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705304357495568e-05,
+      "loss": 0.6118,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970402526075846e-05,
+      "loss": 0.5837,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702743435829268e-05,
+      "loss": 0.6565,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701458883068368e-05,
+      "loss": 0.6646,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700171602836895e-05,
+      "loss": 0.672,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969888159549676e-05,
+      "loss": 0.6196,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697588861410633e-05,
+      "loss": 0.7351,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696293400941955e-05,
+      "loss": 0.5791,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694995214454932e-05,
+      "loss": 0.7461,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693694302314534e-05,
+      "loss": 0.624,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692390664886502e-05,
+      "loss": 0.6958,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969108430253734e-05,
+      "loss": 0.6611,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968977521563432e-05,
+      "loss": 0.6667,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968846340454548e-05,
+      "loss": 0.7275,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687148869639624e-05,
+      "loss": 0.625,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.832,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968451162985589e-05,
+      "loss": 0.6904,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968318892571945e-05,
+      "loss": 0.7156,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968186349924886e-05,
+      "loss": 0.5547,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968053535081675e-05,
+      "loss": 0.6235,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679204480796518e-05,
+      "loss": 0.5781,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677870889562324e-05,
+      "loss": 0.6487,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967653457748909e-05,
+      "loss": 0.6238,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675195544952515e-05,
+      "loss": 0.6565,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967385379232905e-05,
+      "loss": 0.6213,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672509319995918e-05,
+      "loss": 0.7551,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671162128331104e-05,
+      "loss": 0.8389,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966981221771335e-05,
+      "loss": 0.5771,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668459588522182e-05,
+      "loss": 0.5276,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667104241137874e-05,
+      "loss": 0.6392,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966574617594147e-05,
+      "loss": 0.4895,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664385393314765e-05,
+      "loss": 0.751,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663021893640348e-05,
+      "loss": 0.7866,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661655677301543e-05,
+      "loss": 0.6592,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966028674468245e-05,
+      "loss": 0.6528,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658915096167933e-05,
+      "loss": 0.6804,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965754073214362e-05,
+      "loss": 0.5654,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656163652995892e-05,
+      "loss": 0.7065,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654783859111905e-05,
+      "loss": 0.4983,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653401350879577e-05,
+      "loss": 0.6812,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652016128687586e-05,
+      "loss": 0.7178,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650628192925372e-05,
+      "loss": 0.6484,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964923754398314e-05,
+      "loss": 0.5308,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964784418225186e-05,
+      "loss": 0.6245,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646448108123258e-05,
+      "loss": 0.6592,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645049321989827e-05,
+      "loss": 0.8037,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643647824244825e-05,
+      "loss": 0.6877,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642243615282267e-05,
+      "loss": 0.7021,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640836695496935e-05,
+      "loss": 0.7109,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963942706528437e-05,
+      "loss": 0.718,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638014725040876e-05,
+      "loss": 0.6133,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636599675163516e-05,
+      "loss": 0.6768,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635181916050116e-05,
+      "loss": 0.7422,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633761448099276e-05,
+      "loss": 0.5977,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632338271710333e-05,
+      "loss": 0.626,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963091238728341e-05,
+      "loss": 0.6704,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629483795219375e-05,
+      "loss": 0.6284,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628052495919862e-05,
+      "loss": 0.6951,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962661848978727e-05,
+      "loss": 0.7832,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625181777224752e-05,
+      "loss": 0.5801,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962374235863623e-05,
+      "loss": 0.6885,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962230023442638e-05,
+      "loss": 0.6726,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962085540500064e-05,
+      "loss": 0.5991,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619407870765214e-05,
+      "loss": 0.6802,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961795763212706e-05,
+      "loss": 0.7056,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616504689493898e-05,
+      "loss": 0.6345,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615049043274207e-05,
+      "loss": 0.5774,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613590693877228e-05,
+      "loss": 0.6133,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612129641712968e-05,
+      "loss": 0.7983,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610665887192178e-05,
+      "loss": 0.5208,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609199430726383e-05,
+      "loss": 0.7803,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960773027272787e-05,
+      "loss": 0.6841,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606258413609662e-05,
+      "loss": 0.7144,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604783853785574e-05,
+      "loss": 0.4932,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603306593670152e-05,
+      "loss": 0.5767,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601826633678725e-05,
+      "loss": 0.5371,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960034397422736e-05,
+      "loss": 0.5867,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598858615732893e-05,
+      "loss": 0.7085,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597370558612924e-05,
+      "loss": 0.6541,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95958798032858e-05,
+      "loss": 0.6477,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594386350170632e-05,
+      "loss": 0.709,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592890199687295e-05,
+      "loss": 0.6802,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591391352256414e-05,
+      "loss": 0.853,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589889808299378e-05,
+      "loss": 0.8789,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588385568238323e-05,
+      "loss": 0.6443,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958687863249616e-05,
+      "loss": 0.6987,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585369001496547e-05,
+      "loss": 0.5542,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583856675663905e-05,
+      "loss": 0.7744,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95823416554234e-05,
+      "loss": 0.7734,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580823941200978e-05,
+      "loss": 0.6377,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957930353342332e-05,
+      "loss": 0.6187,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957778043251788e-05,
+      "loss": 0.6982,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576254638912854e-05,
+      "loss": 0.6619,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574726153037216e-05,
+      "loss": 0.7668,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.6692,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957166110619371e-05,
+      "loss": 0.6135,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570124546087555e-05,
+      "loss": 0.6215,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95685852954342e-05,
+      "loss": 0.5625,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567043354666385e-05,
+      "loss": 0.645,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565498724217617e-05,
+      "loss": 0.624,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956395140452215e-05,
+      "loss": 0.6609,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562401396015e-05,
+      "loss": 0.644,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560848699131937e-05,
+      "loss": 0.5483,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559293314309483e-05,
+      "loss": 0.6067,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557735241984926e-05,
+      "loss": 0.6177,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556174482596296e-05,
+      "loss": 0.7058,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554611036582388e-05,
+      "loss": 0.625,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955304490438275e-05,
+      "loss": 0.6299,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551476086437685e-05,
+      "loss": 0.623,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549904583188254e-05,
+      "loss": 0.6621,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954833039507626e-05,
+      "loss": 0.531,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954675352254428e-05,
+      "loss": 0.5635,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954517396603563e-05,
+      "loss": 0.5541,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543591725994395e-05,
+      "loss": 0.5781,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95420068028654e-05,
+      "loss": 0.6887,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540419197094232e-05,
+      "loss": 0.5938,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953882890912723e-05,
+      "loss": 0.6031,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537235939411488e-05,
+      "loss": 0.6479,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535640288394857e-05,
+      "loss": 0.6331,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534041956525934e-05,
+      "loss": 0.5525,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953244094425408e-05,
+      "loss": 0.5854,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95308372520294e-05,
+      "loss": 0.6943,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529230880302753e-05,
+      "loss": 0.6516,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527621829525766e-05,
+      "loss": 0.6337,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526010100150797e-05,
+      "loss": 0.5518,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524395692630972e-05,
+      "loss": 0.6816,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952277860742017e-05,
+      "loss": 0.5298,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952115884497301e-05,
+      "loss": 0.8613,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951953640574488e-05,
+      "loss": 0.6365,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951791129019191e-05,
+      "loss": 0.6309,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516283498770985e-05,
+      "loss": 0.6206,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951465303193974e-05,
+      "loss": 0.5789,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513019890156573e-05,
+      "loss": 0.583,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511384073880623e-05,
+      "loss": 0.6301,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509745583571776e-05,
+      "loss": 0.59,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508104419690685e-05,
+      "loss": 0.7102,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506460582698746e-05,
+      "loss": 0.5481,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504814073058108e-05,
+      "loss": 0.6509,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950316489123167e-05,
+      "loss": 0.6187,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501513037683085e-05,
+      "loss": 0.5862,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499858512876753e-05,
+      "loss": 0.5173,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949820131727783e-05,
+      "loss": 0.6379,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496541451352217e-05,
+      "loss": 0.6873,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949487891556657e-05,
+      "loss": 0.6123,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493213710388302e-05,
+      "loss": 0.6133,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949154583628556e-05,
+      "loss": 0.6204,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948987529372725e-05,
+      "loss": 0.4723,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948820208318304e-05,
+      "loss": 0.79,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486526205123324e-05,
+      "loss": 0.6375,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484847660019266e-05,
+      "loss": 0.6536,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483166448342773e-05,
+      "loss": 0.6016,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481482570566496e-05,
+      "loss": 0.7698,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479796027163847e-05,
+      "loss": 0.752,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478106818608973e-05,
+      "loss": 0.7817,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947641494537679e-05,
+      "loss": 0.791,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947472040794294e-05,
+      "loss": 0.5964,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473023206783834e-05,
+      "loss": 0.6479,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471323342376617e-05,
+      "loss": 0.7051,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469620815199195e-05,
+      "loss": 0.5771,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467915625730216e-05,
+      "loss": 0.7144,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466207774449074e-05,
+      "loss": 0.7402,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946449726183592e-05,
+      "loss": 0.627,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946278408837164e-05,
+      "loss": 0.7473,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461068254537885e-05,
+      "loss": 0.6377,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945934976081704e-05,
+      "loss": 0.6606,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457628607692243e-05,
+      "loss": 0.5472,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945590479564738e-05,
+      "loss": 0.7505,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454178325167086e-05,
+      "loss": 0.6182,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452449196736735e-05,
+      "loss": 0.6572,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450717410842462e-05,
+      "loss": 0.5701,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448982967971136e-05,
+      "loss": 0.6897,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447245868610385e-05,
+      "loss": 0.6504,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944550611324857e-05,
+      "loss": 0.6011,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 0.748,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442018636478968e-05,
+      "loss": 0.6677,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440270916051657e-05,
+      "loss": 0.7163,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438520541584218e-05,
+      "loss": 0.7122,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943676751356876e-05,
+      "loss": 0.6201,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435011832498132e-05,
+      "loss": 0.6799,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433253498865926e-05,
+      "loss": 0.5637,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943149251316647e-05,
+      "loss": 0.7224,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942972887589486e-05,
+      "loss": 0.7412,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427962587546925e-05,
+      "loss": 0.6248,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426193648619235e-05,
+      "loss": 0.7402,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942442205960911e-05,
+      "loss": 0.7036,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422647821014617e-05,
+      "loss": 0.6904,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420870933334563e-05,
+      "loss": 0.6719,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419091397068507e-05,
+      "loss": 0.7659,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417309212716745e-05,
+      "loss": 0.686,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941552438078032e-05,
+      "loss": 0.7231,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413736901761025e-05,
+      "loss": 0.7317,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411946776161388e-05,
+      "loss": 0.5693,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410154004484684e-05,
+      "loss": 0.6355,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940835858723494e-05,
+      "loss": 0.7239,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406560524916914e-05,
+      "loss": 0.7263,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940475981803611e-05,
+      "loss": 0.4622,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402956467098792e-05,
+      "loss": 0.6785,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401150472611946e-05,
+      "loss": 0.5598,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939934183508331e-05,
+      "loss": 0.5747,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397530555021365e-05,
+      "loss": 0.7261,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395716632935336e-05,
+      "loss": 0.6704,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939390006933519e-05,
+      "loss": 0.811,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392080864731632e-05,
+      "loss": 0.6047,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939025901963612e-05,
+      "loss": 0.5718,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938843453456084e-05,
+      "loss": 0.7227,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938660741001874e-05,
+      "loss": 0.6836,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938477764652349e-05,
+      "loss": 0.5972,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382945244589505e-05,
+      "loss": 0.7314,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938111020473196e-05,
+      "loss": 0.5692,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937927252746675e-05,
+      "loss": 0.6724,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377432213310522e-05,
+      "loss": 0.6428,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937558926278066e-05,
+      "loss": 0.6458,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373743676395295e-05,
+      "loss": 0.6189,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371895454673296e-05,
+      "loss": 0.7524,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937004459813427e-05,
+      "loss": 0.5935,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368191107298567e-05,
+      "loss": 0.7205,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936633498268728e-05,
+      "loss": 0.6592,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936447622482224e-05,
+      "loss": 0.667,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362614834226017e-05,
+      "loss": 0.6743,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360750811421923e-05,
+      "loss": 0.606,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358884156934014e-05,
+      "loss": 0.6436,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357014871287078e-05,
+      "loss": 0.5083,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355142955006645e-05,
+      "loss": 0.4908,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935326840861899e-05,
+      "loss": 0.4664,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935139123265112e-05,
+      "loss": 0.5391,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349511427630785e-05,
+      "loss": 0.6235,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347628994086477e-05,
+      "loss": 0.6846,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934574393254742e-05,
+      "loss": 0.7249,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343856243543584e-05,
+      "loss": 0.687,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341965927605672e-05,
+      "loss": 0.6753,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340072985265128e-05,
+      "loss": 0.53,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338177417054135e-05,
+      "loss": 0.7026,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336279223505617e-05,
+      "loss": 0.7036,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334378405153225e-05,
+      "loss": 0.644,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332474962531362e-05,
+      "loss": 0.5518,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933056889617516e-05,
+      "loss": 0.5669,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328660206620488e-05,
+      "loss": 0.6064,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326748894403963e-05,
+      "loss": 0.6091,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324834960062922e-05,
+      "loss": 0.6909,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932291840413546e-05,
+      "loss": 0.6562,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320999227160387e-05,
+      "loss": 0.7319,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931907742967727e-05,
+      "loss": 0.574,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317153012226396e-05,
+      "loss": 0.6995,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315225975348802e-05,
+      "loss": 0.6458,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313296319586252e-05,
+      "loss": 0.5754,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311364045481252e-05,
+      "loss": 0.7986,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309429153577042e-05,
+      "loss": 0.6685,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307491644417597e-05,
+      "loss": 0.667,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930555151854763e-05,
+      "loss": 0.606,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303608776512587e-05,
+      "loss": 0.7729,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301663418858654e-05,
+      "loss": 0.6392,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299715446132747e-05,
+      "loss": 0.728,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.584,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295811657656353e-05,
+      "loss": 0.8018,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293855843003383e-05,
+      "loss": 0.6248,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291897415473465e-05,
+      "loss": 0.7141,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928993637561719e-05,
+      "loss": 0.7102,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287972723985883e-05,
+      "loss": 0.772,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928600646113161e-05,
+      "loss": 0.6016,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284037587607157e-05,
+      "loss": 0.7087,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282066103966067e-05,
+      "loss": 0.6321,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280092010762597e-05,
+      "loss": 0.6785,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278115308551743e-05,
+      "loss": 0.748,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276135997889233e-05,
+      "loss": 0.8101,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927415407933154e-05,
+      "loss": 0.6826,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272169553435854e-05,
+      "loss": 0.5884,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270182420760104e-05,
+      "loss": 0.6904,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268192681862957e-05,
+      "loss": 0.6682,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926620033730381e-05,
+      "loss": 0.6479,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264205387642783e-05,
+      "loss": 0.6885,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262207833440748e-05,
+      "loss": 0.6848,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260207675259292e-05,
+      "loss": 0.7622,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258204913660737e-05,
+      "loss": 0.5205,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256199549208144e-05,
+      "loss": 0.5278,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92541915824653e-05,
+      "loss": 0.6562,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252181013996728e-05,
+      "loss": 0.5486,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925016784436768e-05,
+      "loss": 0.5457,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924815207414413e-05,
+      "loss": 0.5149,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246133703892804e-05,
+      "loss": 0.6033,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244112734181142e-05,
+      "loss": 0.6953,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924208916557732e-05,
+      "loss": 0.6069,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924006299865024e-05,
+      "loss": 0.6802,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238034233969546e-05,
+      "loss": 0.5359,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236002872105603e-05,
+      "loss": 0.5134,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233968913629512e-05,
+      "loss": 0.6624,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923193235911309e-05,
+      "loss": 0.7427,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229893209128905e-05,
+      "loss": 0.5859,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227851464250246e-05,
+      "loss": 0.5247,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225807125051117e-05,
+      "loss": 0.6746,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223760192106274e-05,
+      "loss": 0.5869,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221710665991188e-05,
+      "loss": 0.6411,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921965854728207e-05,
+      "loss": 0.6228,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217603836555843e-05,
+      "loss": 0.5886,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215546534390178e-05,
+      "loss": 0.7119,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213486641363458e-05,
+      "loss": 0.491,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211424158054806e-05,
+      "loss": 0.7231,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209359085044067e-05,
+      "loss": 0.7175,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920729142291182e-05,
+      "loss": 0.7229,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205221172239367e-05,
+      "loss": 0.6409,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203148333608735e-05,
+      "loss": 0.6326,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201072907602683e-05,
+      "loss": 0.5399,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.91989948948047e-05,
+      "loss": 0.699,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196914295799e-05,
+      "loss": 0.6128,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919483111117052e-05,
+      "loss": 0.6331,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192745341504924e-05,
+      "loss": 0.5571,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190656987388613e-05,
+      "loss": 0.5601,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91885660494087e-05,
+      "loss": 0.5867,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918647252815304e-05,
+      "loss": 0.5515,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91843764242102e-05,
+      "loss": 0.6211,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918227773816948e-05,
+      "loss": 0.7051,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180176470620905e-05,
+      "loss": 0.5403,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917807262215523e-05,
+      "loss": 0.6111,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917596619336393e-05,
+      "loss": 0.6133,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173857184839203e-05,
+      "loss": 0.6494,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171745597173977e-05,
+      "loss": 0.6472,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916963143096191e-05,
+      "loss": 0.6292,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916751468679737e-05,
+      "loss": 0.7437,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165395365275468e-05,
+      "loss": 0.6523,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916327346699202e-05,
+      "loss": 0.7017,
+      "step": 961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161148992543595e-05,
+      "loss": 0.5033,
+      "step": 962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915902194252745e-05,
+      "loss": 0.582,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915689231754159e-05,
+      "loss": 0.6135,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154760118184744e-05,
+      "loss": 0.6487,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152625345056352e-05,
+      "loss": 0.5913,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150487998756586e-05,
+      "loss": 0.6052,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148348079886343e-05,
+      "loss": 0.6118,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146205589047237e-05,
+      "loss": 0.6729,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144060526841614e-05,
+      "loss": 0.6851,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914191289387253e-05,
+      "loss": 0.7068,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139762690743776e-05,
+      "loss": 0.5604,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137609918059857e-05,
+      "loss": 0.5044,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.5613,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133296666448184e-05,
+      "loss": 0.5178,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913113618873305e-05,
+      "loss": 0.5288,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128973143888013e-05,
+      "loss": 0.5564,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912680753252119e-05,
+      "loss": 0.6846,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912463935524142e-05,
+      "loss": 0.5049,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122468612658268e-05,
+      "loss": 0.7266,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120295305382012e-05,
+      "loss": 0.6394,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118119434023658e-05,
+      "loss": 0.7637,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115940999194932e-05,
+      "loss": 0.7397,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911376000150828e-05,
+      "loss": 0.582,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111576441576866e-05,
+      "loss": 0.6589,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109390320014583e-05,
+      "loss": 0.5562,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910720163743603e-05,
+      "loss": 0.647,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105010394456537e-05,
+      "loss": 0.6863,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102816591692148e-05,
+      "loss": 0.7314,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100620229759635e-05,
+      "loss": 0.7129,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098421309276477e-05,
+      "loss": 0.7075,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096219830860887e-05,
+      "loss": 0.5981,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909401579513178e-05,
+      "loss": 0.6067,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091809202708805e-05,
+      "loss": 0.6707,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089600054212325e-05,
+      "loss": 0.5886,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087388350263418e-05,
+      "loss": 0.5994,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085174091483882e-05,
+      "loss": 0.6147,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082957278496236e-05,
+      "loss": 0.6169,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080737911923714e-05,
+      "loss": 0.5913,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907851599239027e-05,
+      "loss": 0.5771,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907629152052058e-05,
+      "loss": 0.5564,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074064496940023e-05,
+      "loss": 0.5408,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071834922274707e-05,
+      "loss": 0.6042,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069602797151465e-05,
+      "loss": 0.8176,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067368122197824e-05,
+      "loss": 0.5361,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906513089804205e-05,
+      "loss": 0.6501,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062891125313116e-05,
+      "loss": 0.7056,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060648804640706e-05,
+      "loss": 0.6714,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058403936655235e-05,
+      "loss": 0.5901,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905615652198782e-05,
+      "loss": 0.7271,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90539065612703e-05,
+      "loss": 0.6833,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905165405513523e-05,
+      "loss": 0.5686,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904939900421588e-05,
+      "loss": 0.5225,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047141409146243e-05,
+      "loss": 0.6272,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044881270561008e-05,
+      "loss": 0.6106,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042618589095594e-05,
+      "loss": 0.6592,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904035336538614e-05,
+      "loss": 0.7549,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038085600069486e-05,
+      "loss": 0.6316,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035815293783186e-05,
+      "loss": 0.7136,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033542447165527e-05,
+      "loss": 0.6133,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031267060855487e-05,
+      "loss": 0.6589,
+      "step": 1021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028989135492773e-05,
+      "loss": 0.7122,
+      "step": 1022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90267086717178e-05,
+      "loss": 0.6587,
+      "step": 1023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90244256701717e-05,
+      "loss": 0.6191,
+      "step": 1024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022140131496317e-05,
+      "loss": 0.5525,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019852056334206e-05,
+      "loss": 0.5378,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017561445328638e-05,
+      "loss": 0.6211,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015268299123593e-05,
+      "loss": 0.6567,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901297261836377e-05,
+      "loss": 0.6428,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901067440369458e-05,
+      "loss": 0.8315,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900837365576214e-05,
+      "loss": 0.5625,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006070375213284e-05,
+      "loss": 0.6165,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900376456269555e-05,
+      "loss": 0.6897,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001456218857207e-05,
+      "loss": 0.7715,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899914534434722e-05,
+      "loss": 0.5991,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996831939815258e-05,
+      "loss": 0.6267,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994516005911724e-05,
+      "loss": 0.7051,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992197543287714e-05,
+      "loss": 0.644,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989876552595045e-05,
+      "loss": 0.574,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898755303448624e-05,
+      "loss": 0.6628,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985226989614532e-05,
+      "loss": 0.4556,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898289841863387e-05,
+      "loss": 0.7681,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89805673221989e-05,
+      "loss": 0.6619,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978233700964996e-05,
+      "loss": 0.644,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975897555588233e-05,
+      "loss": 0.5889,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973558886725385e-05,
+      "loss": 0.5752,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971217695033958e-05,
+      "loss": 0.6598,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968873981172148e-05,
+      "loss": 0.7075,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966527745798872e-05,
+      "loss": 0.6846,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964178989573746e-05,
+      "loss": 0.6221,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961827713157105e-05,
+      "loss": 0.5708,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959473917209982e-05,
+      "loss": 0.696,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.707,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954758769372e-05,
+      "loss": 0.6418,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952397418806754e-05,
+      "loss": 0.6753,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950033551362267e-05,
+      "loss": 0.749,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947667167703116e-05,
+      "loss": 0.6841,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894529826849458e-05,
+      "loss": 0.6838,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894292685440266e-05,
+      "loss": 0.6646,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940552926094055e-05,
+      "loss": 0.676,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938176484236168e-05,
+      "loss": 0.627,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935797529497115e-05,
+      "loss": 0.6687,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933416062545717e-05,
+      "loss": 0.4878,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931032084051496e-05,
+      "loss": 0.666,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892864559468469e-05,
+      "loss": 0.6169,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926256595116233e-05,
+      "loss": 0.6011,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892386508601777e-05,
+      "loss": 0.6799,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892147106806165e-05,
+      "loss": 0.574,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919074541920928e-05,
+      "loss": 0.6672,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916675508269363e-05,
+      "loss": 0.6426,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914273967781423e-05,
+      "loss": 0.7205,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891186992113228e-05,
+      "loss": 0.6345,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89094633689978e-05,
+      "loss": 0.6367,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907054312054566e-05,
+      "loss": 0.738,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904642750979862e-05,
+      "loss": 0.5771,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902228686451672e-05,
+      "loss": 0.522,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899812119148687e-05,
+      "loss": 0.5614,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897393049750308e-05,
+      "loss": 0.7166,
+      "step": 1078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894971478936625e-05,
+      "loss": 0.6748,
+      "step": 1079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889254740738844e-05,
+      "loss": 0.623,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889012083578726e-05,
+      "loss": 0.5447,
+      "step": 1081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888769176481529e-05,
+      "loss": 0.6562,
+      "step": 1082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885260195155436e-05,
+      "loss": 0.5325,
+      "step": 1083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888282612749132e-05,
+      "loss": 0.5525,
+      "step": 1084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880389562507248e-05,
+      "loss": 0.6045,
+      "step": 1085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877950500888242e-05,
+      "loss": 0.5972,
+      "step": 1086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875508943320015e-05,
+      "loss": 0.6494,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887306489048899e-05,
+      "loss": 0.5667,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887061834308229e-05,
+      "loss": 0.6133,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868169301787734e-05,
+      "loss": 0.6882,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886571776729385e-05,
+      "loss": 0.582,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886326374028986e-05,
+      "loss": 0.8442,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886080722146569e-05,
+      "loss": 0.6006,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885834821151197e-05,
+      "loss": 0.7593,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885588671112002e-05,
+      "loss": 0.6099,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853422720981873e-05,
+      "loss": 0.7153,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850956241790257e-05,
+      "loss": 0.7097,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884848727423859e-05,
+      "loss": 0.6753,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884601581902101e-05,
+      "loss": 0.7786,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843541876832328e-05,
+      "loss": 0.6338,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884106544836808e-05,
+      "loss": 0.6191,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838586534324488e-05,
+      "loss": 0.5532,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836105135398472e-05,
+      "loss": 0.6228,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883362125228765e-05,
+      "loss": 0.5452,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883113488569035e-05,
+      "loss": 0.6167,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828646036305588e-05,
+      "loss": 0.6802,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882615470483307e-05,
+      "loss": 0.6035,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823660891973223e-05,
+      "loss": 0.5544,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821164598427148e-05,
+      "loss": 0.5459,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818665824896657e-05,
+      "loss": 0.6362,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816164572084256e-05,
+      "loss": 0.6584,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881366084069315e-05,
+      "loss": 0.7593,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811154631427237e-05,
+      "loss": 0.6418,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880864594499111e-05,
+      "loss": 0.5767,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806134782090066e-05,
+      "loss": 0.5439,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803621143430093e-05,
+      "loss": 0.7729,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801105029717875e-05,
+      "loss": 0.7471,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798586441660795e-05,
+      "loss": 0.646,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796065379966927e-05,
+      "loss": 0.5894,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793541845345045e-05,
+      "loss": 0.5659,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879101583850461e-05,
+      "loss": 0.6079,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788487360155793e-05,
+      "loss": 0.709,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878595641100945e-05,
+      "loss": 0.7041,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783422991777126e-05,
+      "loss": 0.7214,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878088710317107e-05,
+      "loss": 0.5848,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778348745904228e-05,
+      "loss": 0.6804,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775807920690224e-05,
+      "loss": 0.6821,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773264628243395e-05,
+      "loss": 0.6687,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770718869278762e-05,
+      "loss": 0.7646,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768170644512037e-05,
+      "loss": 0.7227,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765619954659627e-05,
+      "loss": 0.6262,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.8022,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760511182566864e-05,
+      "loss": 0.6271,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757953101762786e-05,
+      "loss": 0.4587,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755392558745593e-05,
+      "loss": 0.6245,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875282955423515e-05,
+      "loss": 0.5776,
+      "step": 1136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750264088952023e-05,
+      "loss": 0.8135,
+      "step": 1137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747696163617467e-05,
+      "loss": 0.5815,
+      "step": 1138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874512577895343e-05,
+      "loss": 0.6772,
+      "step": 1139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874255293568255e-05,
+      "loss": 0.604,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873997763452816e-05,
+      "loss": 0.6099,
+      "step": 1141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737399876214274e-05,
+      "loss": 0.5974,
+      "step": 1142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734819661465607e-05,
+      "loss": 0.5659,
+      "step": 1143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873223699100756e-05,
+      "loss": 0.752,
+      "step": 1144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872965186556623e-05,
+      "loss": 0.5317,
+      "step": 1145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727064285868396e-05,
+      "loss": 0.5889,
+      "step": 1146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872447425264153e-05,
+      "loss": 0.7234,
+      "step": 1147
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87218817666138e-05,
+      "loss": 0.749,
+      "step": 1148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871928682851405e-05,
+      "loss": 0.7964,
+      "step": 1149
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716689439071827e-05,
+      "loss": 0.6687,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871408959901736e-05,
+      "loss": 0.5408,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711487309081573e-05,
+      "loss": 0.6211,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870888256999607e-05,
+      "loss": 0.6377,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706275382493146e-05,
+      "loss": 0.7019,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870366574730579e-05,
+      "loss": 0.5815,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701053665167673e-05,
+      "loss": 0.6715,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698439136813162e-05,
+      "loss": 0.5344,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86958221629773e-05,
+      "loss": 0.553,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869320274439583e-05,
+      "loss": 0.551,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690580881805166e-05,
+      "loss": 0.7566,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868795657594243e-05,
+      "loss": 0.5913,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685329827545407e-05,
+      "loss": 0.6836,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682700637352594e-05,
+      "loss": 0.583,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868006900610316e-05,
+      "loss": 0.6958,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677434934536958e-05,
+      "loss": 0.6199,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674798423394535e-05,
+      "loss": 0.5657,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867215947341712e-05,
+      "loss": 0.6248,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669518085346624e-05,
+      "loss": 0.5989,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666874259925654e-05,
+      "loss": 0.6365,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866422799789749e-05,
+      "loss": 0.6248,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866157930000611e-05,
+      "loss": 0.5466,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658928166996163e-05,
+      "loss": 0.7827,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865627459961299e-05,
+      "loss": 0.582,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653618598602615e-05,
+      "loss": 0.7451,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650960164711752e-05,
+      "loss": 0.5344,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648299298687788e-05,
+      "loss": 0.6501,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86456360012788e-05,
+      "loss": 0.5696,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642970273233553e-05,
+      "loss": 0.6121,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640302115301486e-05,
+      "loss": 0.6814,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863763152823273e-05,
+      "loss": 0.6646,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634958512778085e-05,
+      "loss": 0.6968,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632283069689054e-05,
+      "loss": 0.5396,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629605199717805e-05,
+      "loss": 0.6589,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86269249036172e-05,
+      "loss": 0.7185,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624242182140776e-05,
+      "loss": 0.6943,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862155703604275e-05,
+      "loss": 0.6692,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861886946607803e-05,
+      "loss": 0.7417,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86161794730022e-05,
+      "loss": 0.5289,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613487057571523e-05,
+      "loss": 0.6145,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610792220542944e-05,
+      "loss": 0.7822,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608094962674092e-05,
+      "loss": 0.5449,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605395284723277e-05,
+      "loss": 0.6335,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602693187449484e-05,
+      "loss": 0.5729,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599988671612382e-05,
+      "loss": 0.6748,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597281737972323e-05,
+      "loss": 0.688,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594572387290324e-05,
+      "loss": 0.5864,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591860620328107e-05,
+      "loss": 0.5417,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858914643784805e-05,
+      "loss": 0.719,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858642984061322e-05,
+      "loss": 0.551,
+      "step": 1199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583710829387364e-05,
+      "loss": 0.5864,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580989404934902e-05,
+      "loss": 0.5723,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857826556802094e-05,
+      "loss": 0.5869,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575539319411254e-05,
+      "loss": 0.5742,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572810659872307e-05,
+      "loss": 0.7334,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570079590171232e-05,
+      "loss": 0.6931,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567346111075842e-05,
+      "loss": 0.5249,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856461022335463e-05,
+      "loss": 0.7158,
+      "step": 1207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561871927776763e-05,
+      "loss": 0.584,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559131225112085e-05,
+      "loss": 0.5193,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556388116131122e-05,
+      "loss": 0.522,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 0.6003,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550894682305797e-05,
+      "loss": 0.7092,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548144359005865e-05,
+      "loss": 0.7043,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545391632478494e-05,
+      "loss": 0.5818,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854263650349759e-05,
+      "loss": 0.7573,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539878972837728e-05,
+      "loss": 0.741,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853711904127416e-05,
+      "loss": 0.4805,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534356709582817e-05,
+      "loss": 0.5171,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531591978540297e-05,
+      "loss": 0.5618,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528824848923884e-05,
+      "loss": 0.5901,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526055321511517e-05,
+      "loss": 0.6926,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852328339708184e-05,
+      "loss": 0.5344,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520509076414133e-05,
+      "loss": 0.7166,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517732360288386e-05,
+      "loss": 0.5764,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514953249485236e-05,
+      "loss": 0.6277,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512171744786007e-05,
+      "loss": 0.5942,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850938784697269e-05,
+      "loss": 0.7158,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850660155682795e-05,
+      "loss": 0.7407,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503812875135122e-05,
+      "loss": 0.6035,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501021802678228e-05,
+      "loss": 0.6655,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849822834024194e-05,
+      "loss": 0.6592,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495432488611616e-05,
+      "loss": 0.4932,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492634248573283e-05,
+      "loss": 0.5632,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489833620913644e-05,
+      "loss": 0.7292,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487030606420055e-05,
+      "loss": 0.7568,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484225205880567e-05,
+      "loss": 0.7017,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481417420083892e-05,
+      "loss": 0.7104,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478607249819404e-05,
+      "loss": 0.6189,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475794695877157e-05,
+      "loss": 0.6733,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847297975904788e-05,
+      "loss": 0.7002,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847016244012296e-05,
+      "loss": 0.5945,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846734273989445e-05,
+      "loss": 0.5991,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464520659155097e-05,
+      "loss": 0.6157,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461696198698295e-05,
+      "loss": 0.6614,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845886935931811e-05,
+      "loss": 0.7786,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456040141809284e-05,
+      "loss": 0.4034,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453208546967226e-05,
+      "loss": 0.6362,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450374575588003e-05,
+      "loss": 0.4629,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844753822846837e-05,
+      "loss": 0.6177,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844469950640573e-05,
+      "loss": 0.731,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844185841019816e-05,
+      "loss": 0.636,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439014940644412e-05,
+      "loss": 0.5728,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84361690985439e-05,
+      "loss": 0.6985,
+      "step": 1253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433320884696704e-05,
+      "loss": 0.532,
+      "step": 1254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843047029990357e-05,
+      "loss": 0.6001,
+      "step": 1255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842761734496591e-05,
+      "loss": 0.635,
+      "step": 1256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842476202068581e-05,
+      "loss": 0.6443,
+      "step": 1257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421904327866015e-05,
+      "loss": 0.5818,
+      "step": 1258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841904426730994e-05,
+      "loss": 0.6992,
+      "step": 1259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841618183982165e-05,
+      "loss": 0.6008,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413317046205903e-05,
+      "loss": 0.6249,
+      "step": 1261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410449887268106e-05,
+      "loss": 0.4667,
+      "step": 1262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407580363814324e-05,
+      "loss": 0.561,
+      "step": 1263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404708476651303e-05,
+      "loss": 0.6421,
+      "step": 1264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401834226586447e-05,
+      "loss": 0.6548,
+      "step": 1265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398957614427817e-05,
+      "loss": 0.574,
+      "step": 1266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396078640984148e-05,
+      "loss": 0.6396,
+      "step": 1267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393197307064836e-05,
+      "loss": 0.5835,
+      "step": 1268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839031361347994e-05,
+      "loss": 0.6187,
+      "step": 1269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387427561040184e-05,
+      "loss": 0.5251,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838453915055695e-05,
+      "loss": 0.5208,
+      "step": 1271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381648382842285e-05,
+      "loss": 0.4966,
+      "step": 1272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378755258708905e-05,
+      "loss": 0.7124,
+      "step": 1273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837585977897018e-05,
+      "loss": 0.6538,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372961944440152e-05,
+      "loss": 0.5315,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837006175593351e-05,
+      "loss": 0.5763,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367159214265617e-05,
+      "loss": 0.5317,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364254320252495e-05,
+      "loss": 0.571,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836134707471083e-05,
+      "loss": 0.7261,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358437478457963e-05,
+      "loss": 0.7332,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835552553231189e-05,
+      "loss": 0.6289,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352611237091296e-05,
+      "loss": 0.6714,
+      "step": 1282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834969459361549e-05,
+      "loss": 0.6626,
+      "step": 1283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346775602704464e-05,
+      "loss": 0.6099,
+      "step": 1284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343854265178863e-05,
+      "loss": 0.623,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834093058185999e-05,
+      "loss": 0.5835,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833800455356981e-05,
+      "loss": 0.6501,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335076181130956e-05,
+      "loss": 0.6118,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332145465366705e-05,
+      "loss": 0.6125,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 0.6426,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326277007158436e-05,
+      "loss": 0.5798,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832333926636428e-05,
+      "loss": 0.6089,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320399185544443e-05,
+      "loss": 0.5908,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317456765525506e-05,
+      "loss": 0.5973,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314512007134703e-05,
+      "loss": 0.5583,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311564911199915e-05,
+      "loss": 0.5054,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308615478549696e-05,
+      "loss": 0.6045,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305663710013247e-05,
+      "loss": 0.6638,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302709606420436e-05,
+      "loss": 0.5696,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829975316860177e-05,
+      "loss": 0.6677,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829679439738843e-05,
+      "loss": 0.593,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293833293612243e-05,
+      "loss": 0.6687,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290869858105694e-05,
+      "loss": 0.6719,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287904091701924e-05,
+      "loss": 0.6582,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284935995234732e-05,
+      "loss": 0.7061,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281965569538567e-05,
+      "loss": 0.5857,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278992815448536e-05,
+      "loss": 0.5266,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276017733800403e-05,
+      "loss": 0.6067,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273040325430575e-05,
+      "loss": 0.5332,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270060591176126e-05,
+      "loss": 0.6641,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826707853187478e-05,
+      "loss": 0.6318,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264094148364916e-05,
+      "loss": 0.7039,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826110744148556e-05,
+      "loss": 0.593,
+      "step": 1313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258118412076396e-05,
+      "loss": 0.6506,
+      "step": 1314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825512706097776e-05,
+      "loss": 0.5566,
+      "step": 1315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825213338903065e-05,
+      "loss": 0.6802,
+      "step": 1316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249137397076695e-05,
+      "loss": 0.5261,
+      "step": 1317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246139085958196e-05,
+      "loss": 0.6797,
+      "step": 1318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243138456518092e-05,
+      "loss": 0.524,
+      "step": 1319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240135509599992e-05,
+      "loss": 0.6519,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237130246048137e-05,
+      "loss": 0.5509,
+      "step": 1321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823412266670743e-05,
+      "loss": 0.6602,
+      "step": 1322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231112772423424e-05,
+      "loss": 0.6628,
+      "step": 1323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822810056404232e-05,
+      "loss": 0.6948,
+      "step": 1324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225086042410968e-05,
+      "loss": 0.6514,
+      "step": 1325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222069208376876e-05,
+      "loss": 0.6743,
+      "step": 1326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219050062788194e-05,
+      "loss": 0.6533,
+      "step": 1327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216028606493733e-05,
+      "loss": 0.6865,
+      "step": 1328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213004840342933e-05,
+      "loss": 0.5042,
+      "step": 1329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820997876518591e-05,
+      "loss": 0.6108,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206950381873403e-05,
+      "loss": 0.604,
+      "step": 1331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820391969125682e-05,
+      "loss": 0.7026,
+      "step": 1332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200886694188206e-05,
+      "loss": 0.6533,
+      "step": 1333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197851391520265e-05,
+      "loss": 0.6272,
+      "step": 1334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194813784106337e-05,
+      "loss": 0.6035,
+      "step": 1335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819177387280042e-05,
+      "loss": 0.5935,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818873165845715e-05,
+      "loss": 0.4647,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818568714193182e-05,
+      "loss": 0.5819,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182640324080366e-05,
+      "loss": 0.7153,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179591205759366e-05,
+      "loss": 0.6721,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176539787826057e-05,
+      "loss": 0.7363,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173486071138313e-05,
+      "loss": 0.5691,
+      "step": 1342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170430056554655e-05,
+      "loss": 0.656,
+      "step": 1343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816737174493425e-05,
+      "loss": 0.7295,
+      "step": 1344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816431113713692e-05,
+      "loss": 0.6421,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161248234023117e-05,
+      "loss": 0.5891,
+      "step": 1346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158183036453952e-05,
+      "loss": 0.6414,
+      "step": 1347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155115545291174e-05,
+      "loss": 0.564,
+      "step": 1348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152045761397176e-05,
+      "loss": 0.7104,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148973685634995e-05,
+      "loss": 0.5774,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145899318868324e-05,
+      "loss": 0.5247,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142822661961485e-05,
+      "loss": 0.5933,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813974371577945e-05,
+      "loss": 0.7395,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136662481187838e-05,
+      "loss": 0.6135,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133578959052904e-05,
+      "loss": 0.6516,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130493150241554e-05,
+      "loss": 0.5988,
+      "step": 1356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127405055621336e-05,
+      "loss": 0.53,
+      "step": 1357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124314676060425e-05,
+      "loss": 0.6665,
+      "step": 1358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121222012427666e-05,
+      "loss": 0.5336,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118127065592523e-05,
+      "loss": 0.5769,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115029836425108e-05,
+      "loss": 0.7251,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111930325796184e-05,
+      "loss": 0.6694,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810882853457714e-05,
+      "loss": 0.53,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105724463640025e-05,
+      "loss": 0.5369,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810261811385751e-05,
+      "loss": 0.478,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099509486102918e-05,
+      "loss": 0.75,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096398581250207e-05,
+      "loss": 0.6416,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093285400173978e-05,
+      "loss": 0.7178,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.5098,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087052212852577e-05,
+      "loss": 0.7031,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80839322083598e-05,
+      "loss": 0.5973,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080809931148306e-05,
+      "loss": 0.5286,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077685382095892e-05,
+      "loss": 0.5968,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074558562080994e-05,
+      "loss": 0.5106,
+      "step": 1374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807142947198269e-05,
+      "loss": 0.5498,
+      "step": 1375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806829811268069e-05,
+      "loss": 0.5779,
+      "step": 1376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065164485055344e-05,
+      "loss": 0.6624,
+      "step": 1377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062028589987644e-05,
+      "loss": 0.6006,
+      "step": 1378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058890428359215e-05,
+      "loss": 0.6621,
+      "step": 1379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805575000105232e-05,
+      "loss": 0.7048,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805260730894986e-05,
+      "loss": 0.6704,
+      "step": 1381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049462352935373e-05,
+      "loss": 0.7158,
+      "step": 1382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046315133893034e-05,
+      "loss": 0.6714,
+      "step": 1383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804316565270765e-05,
+      "loss": 0.6597,
+      "step": 1384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040013910264667e-05,
+      "loss": 0.5435,
+      "step": 1385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803685990745017e-05,
+      "loss": 0.5894,
+      "step": 1386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803370364515087e-05,
+      "loss": 0.6797,
+      "step": 1387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030545124254125e-05,
+      "loss": 0.5459,
+      "step": 1388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802738434564792e-05,
+      "loss": 0.5759,
+      "step": 1389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024221310220877e-05,
+      "loss": 0.5881,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021056018862254e-05,
+      "loss": 0.6279,
+      "step": 1391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017888472461943e-05,
+      "loss": 0.5747,
+      "step": 1392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014718671910462e-05,
+      "loss": 0.6094,
+      "step": 1393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011546618098977e-05,
+      "loss": 0.6008,
+      "step": 1394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800837231191928e-05,
+      "loss": 0.6968,
+      "step": 1395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800519575426379e-05,
+      "loss": 0.6287,
+      "step": 1396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002016946025568e-05,
+      "loss": 0.7654,
+      "step": 1397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998835888098308e-05,
+      "loss": 0.5806,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799565258137633e-05,
+      "loss": 0.7192,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799246702675459e-05,
+      "loss": 0.6836,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798927922512868e-05,
+      "loss": 0.5791,
+      "step": 1401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798608917739481e-05,
+      "loss": 0.6995,
+      "step": 1402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982896884449838e-05,
+      "loss": 0.6396,
+      "step": 1403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979702347191245e-05,
+      "loss": 0.6182,
+      "step": 1404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976505566517146e-05,
+      "loss": 0.688,
+      "step": 1405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797330654332628e-05,
+      "loss": 0.6299,
+      "step": 1406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970105278518022e-05,
+      "loss": 0.6155,
+      "step": 1407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796690177299238e-05,
+      "loss": 0.6038,
+      "step": 1408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796369602764999e-05,
+      "loss": 0.5215,
+      "step": 1409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960488043392104e-05,
+      "loss": 0.7288,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795727782112063e-05,
+      "loss": 0.6602,
+      "step": 1411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795406536173808e-05,
+      "loss": 0.5037,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950850666147612e-05,
+      "loss": 0.6733,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947633735253005e-05,
+      "loss": 0.6931,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944414569958667e-05,
+      "loss": 0.74,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941193171169632e-05,
+      "loss": 0.6807,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793796953979157e-05,
+      "loss": 0.5049,
+      "step": 1417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934743676730773e-05,
+      "loss": 0.6187,
+      "step": 1418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793151558289416e-05,
+      "loss": 0.656,
+      "step": 1419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928285259189275e-05,
+      "loss": 0.4575,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925052706524294e-05,
+      "loss": 0.5162,
+      "step": 1421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792181792580802e-05,
+      "loss": 0.6143,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918580917949876e-05,
+      "loss": 0.6672,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791534168385992e-05,
+      "loss": 0.6455,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912100224448827e-05,
+      "loss": 0.5127,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79088565406279e-05,
+      "loss": 0.5635,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790561063330908e-05,
+      "loss": 0.6636,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902362503404907e-05,
+      "loss": 0.636,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789911215182857e-05,
+      "loss": 0.5864,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895859579493878e-05,
+      "loss": 0.5613,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892604787315247e-05,
+      "loss": 0.5441,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788934777620774e-05,
+      "loss": 0.6011,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788608854708703e-05,
+      "loss": 0.7188,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788282710086942e-05,
+      "loss": 0.5854,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879563438471833e-05,
+      "loss": 0.6045,
+      "step": 1435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876297560811816e-05,
+      "loss": 0.5659,
+      "step": 1436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873029468807535e-05,
+      "loss": 0.6104,
+      "step": 1437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869759163377786e-05,
+      "loss": 0.5962,
+      "step": 1438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866486645441987e-05,
+      "loss": 0.7222,
+      "step": 1439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863211915920164e-05,
+      "loss": 0.4587,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785993497573298e-05,
+      "loss": 0.4647,
+      "step": 1441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856655825801722e-05,
+      "loss": 0.5508,
+      "step": 1442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853374467048285e-05,
+      "loss": 0.6294,
+      "step": 1443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850090900395186e-05,
+      "loss": 0.5164,
+      "step": 1444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784680512676558e-05,
+      "loss": 0.6873,
+      "step": 1445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843517147083222e-05,
+      "loss": 0.6858,
+      "step": 1446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784022696227249e-05,
+      "loss": 0.5522,
+      "step": 1447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 0.6072,
+      "step": 1448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833639980966568e-05,
+      "loss": 0.6758,
+      "step": 1449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830343186323235e-05,
+      "loss": 0.4592,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827044190255266e-05,
+      "loss": 0.5037,
+      "step": 1451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823742993690144e-05,
+      "loss": 0.5557,
+      "step": 1452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820439597555964e-05,
+      "loss": 0.6653,
+      "step": 1453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817134002781445e-05,
+      "loss": 0.7043,
+      "step": 1454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813826210295922e-05,
+      "loss": 0.6238,
+      "step": 1455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810516221029347e-05,
+      "loss": 0.6997,
+      "step": 1456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780720403591229e-05,
+      "loss": 0.6006,
+      "step": 1457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803889655875948e-05,
+      "loss": 0.6516,
+      "step": 1458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800573081852124e-05,
+      "loss": 0.7419,
+      "step": 1459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797254314773233e-05,
+      "loss": 0.7444,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779393335557232e-05,
+      "loss": 0.6941,
+      "step": 1461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779061020518304e-05,
+      "loss": 0.5977,
+      "step": 1462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787284864539663e-05,
+      "loss": 0.6238,
+      "step": 1463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778395733457708e-05,
+      "loss": 0.5591,
+      "step": 1464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780627616230785e-05,
+      "loss": 0.6631,
+      "step": 1465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77772957104369e-05,
+      "loss": 0.6506,
+      "step": 1466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773961618132162e-05,
+      "loss": 0.4409,
+      "step": 1467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770625340253915e-05,
+      "loss": 0.6233,
+      "step": 1468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767286877740125e-05,
+      "loss": 0.7295,
+      "step": 1469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763946231529358e-05,
+      "loss": 0.686,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760603402560813e-05,
+      "loss": 0.5345,
+      "step": 1471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757258391774295e-05,
+      "loss": 0.5989,
+      "step": 1472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753911200110215e-05,
+      "loss": 0.6648,
+      "step": 1473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750561828509606e-05,
+      "loss": 0.6194,
+      "step": 1474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774721027791411e-05,
+      "loss": 0.6035,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743856549265986e-05,
+      "loss": 0.6058,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740500643508095e-05,
+      "loss": 0.5183,
+      "step": 1477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773714256158393e-05,
+      "loss": 0.6016,
+      "step": 1478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773378230443757e-05,
+      "loss": 0.7456,
+      "step": 1479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730419873013725e-05,
+      "loss": 0.5974,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727055268257708e-05,
+      "loss": 0.7129,
+      "step": 1481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723688491115444e-05,
+      "loss": 0.6104,
+      "step": 1482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720319542533472e-05,
+      "loss": 0.7073,
+      "step": 1483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771694842345894e-05,
+      "loss": 0.5967,
+      "step": 1484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77135751348396e-05,
+      "loss": 0.7441,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710199677623826e-05,
+      "loss": 0.6467,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770682205276059e-05,
+      "loss": 0.6426,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703442261199478e-05,
+      "loss": 0.4919,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700060303890685e-05,
+      "loss": 0.5625,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769667618178502e-05,
+      "loss": 0.6819,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693289895833895e-05,
+      "loss": 0.6033,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768990144698933e-05,
+      "loss": 0.3718,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768651083620395e-05,
+      "loss": 0.6809,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683118064430997e-05,
+      "loss": 0.5381,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679723132624315e-05,
+      "loss": 0.7139,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676326041738358e-05,
+      "loss": 0.5107,
+      "step": 1496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767292679272818e-05,
+      "loss": 0.5972,
+      "step": 1497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766952538654945e-05,
+      "loss": 0.6785,
+      "step": 1498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766612182415844e-05,
+      "loss": 0.5637,
+      "step": 1499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662716106512028e-05,
+      "loss": 0.6431,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76593082345677e-05,
+      "loss": 0.7793,
+      "step": 1501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765589820928354e-05,
+      "loss": 0.6587,
+      "step": 1502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765248603161825e-05,
+      "loss": 0.5366,
+      "step": 1503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764907170253113e-05,
+      "loss": 0.6487,
+      "step": 1504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764565522298208e-05,
+      "loss": 0.6655,
+      "step": 1505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764223659393162e-05,
+      "loss": 0.5942,
+      "step": 1506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763881581634085e-05,
+      "loss": 0.6655,
+      "step": 1507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635392891171497e-05,
+      "loss": 0.6641,
+      "step": 1508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631967819385883e-05,
+      "loss": 0.5623,
+      "step": 1509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628540601946934e-05,
+      "loss": 0.6074,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625111239818177e-05,
+      "loss": 0.6838,
+      "step": 1511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621679733963742e-05,
+      "loss": 0.624,
+      "step": 1512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618246085348364e-05,
+      "loss": 0.7195,
+      "step": 1513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614810294937384e-05,
+      "loss": 0.6006,
+      "step": 1514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761137236369674e-05,
+      "loss": 0.5603,
+      "step": 1515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607932292592965e-05,
+      "loss": 0.5758,
+      "step": 1516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604490082593214e-05,
+      "loss": 0.6359,
+      "step": 1517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760104573466522e-05,
+      "loss": 0.644,
+      "step": 1518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759759924977733e-05,
+      "loss": 0.6531,
+      "step": 1519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594150628898495e-05,
+      "loss": 0.7642,
+      "step": 1520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759069987299826e-05,
+      "loss": 0.7935,
+      "step": 1521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587246983046766e-05,
+      "loss": 0.7236,
+      "step": 1522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583791960014764e-05,
+      "loss": 0.5648,
+      "step": 1523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580334804873595e-05,
+      "loss": 0.7004,
+      "step": 1524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757687551859521e-05,
+      "loss": 0.6074,
+      "step": 1525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757341410215215e-05,
+      "loss": 0.6438,
+      "step": 1526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 0.6877,
+      "step": 1527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566484882665192e-05,
+      "loss": 0.6587,
+      "step": 1528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563017081569365e-05,
+      "loss": 0.6453,
+      "step": 1529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559547154205034e-05,
+      "loss": 0.6729,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556075101547732e-05,
+      "loss": 0.605,
+      "step": 1531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552600924573588e-05,
+      "loss": 0.5769,
+      "step": 1532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549124624259335e-05,
+      "loss": 0.77,
+      "step": 1533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545646201582304e-05,
+      "loss": 0.4464,
+      "step": 1534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542165657520415e-05,
+      "loss": 0.4402,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538682993052194e-05,
+      "loss": 0.5039,
+      "step": 1536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535198209156754e-05,
+      "loss": 0.6514,
+      "step": 1537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531711306813813e-05,
+      "loss": 0.7939,
+      "step": 1538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528222287003673e-05,
+      "loss": 0.5403,
+      "step": 1539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524731150707243e-05,
+      "loss": 0.603,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752123789890602e-05,
+      "loss": 0.6008,
+      "step": 1541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517742532582098e-05,
+      "loss": 0.5833,
+      "step": 1542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514245052718164e-05,
+      "loss": 0.6877,
+      "step": 1543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510745460297502e-05,
+      "loss": 0.5906,
+      "step": 1544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507243756303987e-05,
+      "loss": 0.6907,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750373994172209e-05,
+      "loss": 0.6003,
+      "step": 1546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500234017536874e-05,
+      "loss": 0.6509,
+      "step": 1547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496725984734e-05,
+      "loss": 0.5413,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493215844299706e-05,
+      "loss": 0.6641,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748970359722084e-05,
+      "loss": 0.6853,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748618924448484e-05,
+      "loss": 0.5793,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482672787079726e-05,
+      "loss": 0.6184,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747915422599412e-05,
+      "loss": 0.563,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747563356221723e-05,
+      "loss": 0.7891,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747211079673885e-05,
+      "loss": 0.7212,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468585930549383e-05,
+      "loss": 0.5481,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74650589646398e-05,
+      "loss": 0.6052,
+      "step": 1557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746152990000168e-05,
+      "loss": 0.5737,
+      "step": 1558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457998737627183e-05,
+      "loss": 0.5999,
+      "step": 1559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454465478509056e-05,
+      "loss": 0.4788,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745093012364065e-05,
+      "loss": 0.647,
+      "step": 1561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447392674015886e-05,
+      "loss": 0.6086,
+      "step": 1562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443853130629295e-05,
+      "loss": 0.6799,
+      "step": 1563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744031149447597e-05,
+      "loss": 0.5579,
+      "step": 1564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743676776655162e-05,
+      "loss": 0.7212,
+      "step": 1565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743322194785252e-05,
+      "loss": 0.6035,
+      "step": 1566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429674039375552e-05,
+      "loss": 0.7026,
+      "step": 1567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426124042118173e-05,
+      "loss": 0.6528,
+      "step": 1568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422571957078423e-05,
+      "loss": 0.7268,
+      "step": 1569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419017785254945e-05,
+      "loss": 0.7378,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415461527646953e-05,
+      "loss": 0.5981,
+      "step": 1571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411903185254257e-05,
+      "loss": 0.73,
+      "step": 1572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408342759077255e-05,
+      "loss": 0.6553,
+      "step": 1573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740478025011692e-05,
+      "loss": 0.6541,
+      "step": 1574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401215659374817e-05,
+      "loss": 0.7161,
+      "step": 1575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73976489878531e-05,
+      "loss": 0.6357,
+      "step": 1576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394080236554495e-05,
+      "loss": 0.6643,
+      "step": 1577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390509406482333e-05,
+      "loss": 0.5244,
+      "step": 1578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386936498640513e-05,
+      "loss": 0.7397,
+      "step": 1579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738336151403352e-05,
+      "loss": 0.6277,
+      "step": 1580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737978445366643e-05,
+      "loss": 0.7122,
+      "step": 1581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73762053185449e-05,
+      "loss": 0.6692,
+      "step": 1582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372624109675164e-05,
+      "loss": 0.6157,
+      "step": 1583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369040828064046e-05,
+      "loss": 0.6196,
+      "step": 1584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365455474718957e-05,
+      "loss": 0.6011,
+      "step": 1585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736186805064787e-05,
+      "loss": 0.7422,
+      "step": 1586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735827855685937e-05,
+      "loss": 0.6567,
+      "step": 1587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73546869943626e-05,
+      "loss": 0.4775,
+      "step": 1588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351093364167292e-05,
+      "loss": 0.6704,
+      "step": 1589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347497667283758e-05,
+      "loss": 0.7178,
+      "step": 1590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343899904722903e-05,
+      "loss": 0.5127,
+      "step": 1591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340300077496194e-05,
+      "loss": 0.5756,
+      "step": 1592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733669818661569e-05,
+      "loss": 0.7336,
+      "step": 1593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733309423309403e-05,
+      "loss": 0.583,
+      "step": 1594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329488217944425e-05,
+      "loss": 0.6172,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325880142180675e-05,
+      "loss": 0.5981,
+      "step": 1596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322270006817158e-05,
+      "loss": 0.6836,
+      "step": 1597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731865781286882e-05,
+      "loss": 0.6877,
+      "step": 1598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315043561351198e-05,
+      "loss": 0.7332,
+      "step": 1599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311427253280407e-05,
+      "loss": 0.5981,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307808889673134e-05,
+      "loss": 0.6255,
+      "step": 1601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304188471546647e-05,
+      "loss": 0.843,
+      "step": 1602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730056599991879e-05,
+      "loss": 0.53,
+      "step": 1603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296941475807987e-05,
+      "loss": 0.8032,
+      "step": 1604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293314900233234e-05,
+      "loss": 0.6626,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 0.6189,
+      "step": 1606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728605559877078e-05,
+      "loss": 0.6304,
+      "step": 1607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282422874923953e-05,
+      "loss": 0.5674,
+      "step": 1608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278788103694944e-05,
+      "loss": 0.3951,
+      "step": 1609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275151286105628e-05,
+      "loss": 0.7461,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727151242317847e-05,
+      "loss": 0.7131,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726787151593649e-05,
+      "loss": 0.5879,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72642285654033e-05,
+      "loss": 0.5692,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260583572603084e-05,
+      "loss": 0.6318,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256936538560586e-05,
+      "loss": 0.554,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725328746430114e-05,
+      "loss": 0.5907,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724963635085065e-05,
+      "loss": 0.6372,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245983199235587e-05,
+      "loss": 0.6282,
+      "step": 1618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7242328010482997e-05,
+      "loss": 0.4929,
+      "step": 1619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238670785620507e-05,
+      "loss": 0.6206,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235011525676304e-05,
+      "loss": 0.6335,
+      "step": 1621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723135023167916e-05,
+      "loss": 0.5728,
+      "step": 1622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72276869046584e-05,
+      "loss": 0.5244,
+      "step": 1623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224021545643944e-05,
+      "loss": 0.5698,
+      "step": 1624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722035415566627e-05,
+      "loss": 0.7053,
+      "step": 1625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721668473575643e-05,
+      "loss": 0.6165,
+      "step": 1626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721301328694604e-05,
+      "loss": 0.5483,
+      "step": 1627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209339810267294e-05,
+      "loss": 0.6624,
+      "step": 1628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720566430675296e-05,
+      "loss": 0.813,
+      "step": 1629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720198677743636e-05,
+      "loss": 0.5869,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198307223351404e-05,
+      "loss": 0.5027,
+      "step": 1631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719462564553256e-05,
+      "loss": 0.6614,
+      "step": 1632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719094204501486e-05,
+      "loss": 0.5442,
+      "step": 1633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187256422833928e-05,
+      "loss": 0.8364,
+      "step": 1634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718356878002593e-05,
+      "loss": 0.7012,
+      "step": 1635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179879117627615e-05,
+      "loss": 0.5981,
+      "step": 1636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176187436676287e-05,
+      "loss": 0.5898,
+      "step": 1637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717249373820984e-05,
+      "loss": 0.6277,
+      "step": 1638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168798023266713e-05,
+      "loss": 0.6982,
+      "step": 1639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716510029288592e-05,
+      "loss": 0.5193,
+      "step": 1640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161400548107043e-05,
+      "loss": 0.6187,
+      "step": 1641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157698789970233e-05,
+      "loss": 0.6233,
+      "step": 1642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153995019516193e-05,
+      "loss": 0.7244,
+      "step": 1643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715028923778621e-05,
+      "loss": 0.5552,
+      "step": 1644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714658144582213e-05,
+      "loss": 0.6448,
+      "step": 1645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714287164466636e-05,
+      "loss": 0.6758,
+      "step": 1646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139159835361875e-05,
+      "loss": 0.6982,
+      "step": 1647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713544601895221e-05,
+      "loss": 0.5454,
+      "step": 1648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131730196481472e-05,
+      "loss": 0.5845,
+      "step": 1649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128012368994327e-05,
+      "loss": 0.5786,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712429253753601e-05,
+      "loss": 0.7891,
+      "step": 1651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120570703152307e-05,
+      "loss": 0.6125,
+      "step": 1652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711684686688958e-05,
+      "loss": 0.6641,
+      "step": 1653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711312102979475e-05,
+      "loss": 0.5186,
+      "step": 1654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71093931929153e-05,
+      "loss": 0.7021,
+      "step": 1655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105663357299277e-05,
+      "loss": 0.634,
+      "step": 1656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101931523995284e-05,
+      "loss": 0.489,
+      "step": 1657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709819769405249e-05,
+      "loss": 0.5288,
+      "step": 1658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094461868520625e-05,
+      "loss": 0.4805,
+      "step": 1659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709072404844998e-05,
+      "loss": 0.5618,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708698423489141e-05,
+      "loss": 0.6355,
+      "step": 1661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083242428896323e-05,
+      "loss": 0.6464,
+      "step": 1662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079498631516692e-05,
+      "loss": 0.5967,
+      "step": 1663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707575284380505e-05,
+      "loss": 0.5962,
+      "step": 1664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7072005066814486e-05,
+      "loss": 0.5742,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706825530159866e-05,
+      "loss": 0.7207,
+      "step": 1666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064503549211766e-05,
+      "loss": 0.635,
+      "step": 1667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060749810708585e-05,
+      "loss": 0.7764,
+      "step": 1668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056994087144443e-05,
+      "loss": 0.4749,
+      "step": 1669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053236379575223e-05,
+      "loss": 0.5615,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049476689057367e-05,
+      "loss": 0.7981,
+      "step": 1671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045715016647876e-05,
+      "loss": 0.5262,
+      "step": 1672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704195136340431e-05,
+      "loss": 0.6565,
+      "step": 1673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703818573038478e-05,
+      "loss": 0.593,
+      "step": 1674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034418118647954e-05,
+      "loss": 0.6519,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030648529253067e-05,
+      "loss": 0.614,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70268769632599e-05,
+      "loss": 0.5295,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023103421728788e-05,
+      "loss": 0.7227,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019327905720627e-05,
+      "loss": 0.5952,
+      "step": 1679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701555041629687e-05,
+      "loss": 0.4697,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011770954519517e-05,
+      "loss": 0.6067,
+      "step": 1681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007989521451128e-05,
+      "loss": 0.6958,
+      "step": 1682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700420611815482e-05,
+      "loss": 0.7178,
+      "step": 1683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000420745694256e-05,
+      "loss": 0.5916,
+      "step": 1684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 0.5752,
+      "step": 1685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992844097537793e-05,
+      "loss": 0.5774,
+      "step": 1686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989052823972e-05,
+      "loss": 0.55,
+      "step": 1687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985259585502153e-05,
+      "loss": 0.5476,
+      "step": 1688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698146438319468e-05,
+      "loss": 0.5881,
+      "step": 1689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977667218116572e-05,
+      "loss": 0.457,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697386809133536e-05,
+      "loss": 0.585,
+      "step": 1691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970067003919132e-05,
+      "loss": 0.6074,
+      "step": 1692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696626395693653e-05,
+      "loss": 0.5598,
+      "step": 1693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696245895145674e-05,
+      "loss": 0.8228,
+      "step": 1694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958651988549503e-05,
+      "loss": 0.446,
+      "step": 1695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954843069285113e-05,
+      "loss": 0.5438,
+      "step": 1696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951032194734404e-05,
+      "loss": 0.5564,
+      "step": 1697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694721936596877e-05,
+      "loss": 0.6572,
+      "step": 1698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943404584060156e-05,
+      "loss": 0.6936,
+      "step": 1699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693958785008104e-05,
+      "loss": 0.5928,
+      "step": 1700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935769165104465e-05,
+      "loss": 0.5417,
+      "step": 1701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931948530204018e-05,
+      "loss": 0.5103,
+      "step": 1702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692812594645383e-05,
+      "loss": 0.5413,
+      "step": 1703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924301414928585e-05,
+      "loss": 0.6716,
+      "step": 1704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920474936703515e-05,
+      "loss": 0.4403,
+      "step": 1705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916646512854393e-05,
+      "loss": 0.6755,
+      "step": 1706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912816144457548e-05,
+      "loss": 0.6433,
+      "step": 1707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908983832589843e-05,
+      "loss": 0.6733,
+      "step": 1708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905149578328705e-05,
+      "loss": 0.5906,
+      "step": 1709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690131338275209e-05,
+      "loss": 0.626,
+      "step": 1710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897475246938507e-05,
+      "loss": 0.5559,
+      "step": 1711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893635171967017e-05,
+      "loss": 0.5945,
+      "step": 1712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688979315891721e-05,
+      "loss": 0.7083,
+      "step": 1713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688594920886924e-05,
+      "loss": 0.5586,
+      "step": 1714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688210332290379e-05,
+      "loss": 0.6941,
+      "step": 1715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687825550210209e-05,
+      "loss": 0.6604,
+      "step": 1716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687440574754593e-05,
+      "loss": 0.5688,
+      "step": 1717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687055406031762e-05,
+      "loss": 0.4917,
+      "step": 1718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686670044150003e-05,
+      "loss": 0.6382,
+      "step": 1719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686284489217656e-05,
+      "loss": 0.7263,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858987413431168e-05,
+      "loss": 0.5901,
+      "step": 1721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855128006348343e-05,
+      "loss": 0.729,
+      "step": 1722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685126667201312e-05,
+      "loss": 0.6143,
+      "step": 1723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847403411511075e-05,
+      "loss": 0.6133,
+      "step": 1724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6843538225928332e-05,
+      "loss": 0.6335,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683967111635154e-05,
+      "loss": 0.5771,
+      "step": 1726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835802083867905e-05,
+      "loss": 0.4642,
+      "step": 1727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831931129565167e-05,
+      "loss": 0.54,
+      "step": 1728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828058254531606e-05,
+      "loss": 0.5083,
+      "step": 1729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6824183459856045e-05,
+      "loss": 0.5901,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820306746627842e-05,
+      "loss": 0.6655,
+      "step": 1731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816428115936903e-05,
+      "loss": 0.5769,
+      "step": 1732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812547568873657e-05,
+      "loss": 0.5586,
+      "step": 1733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808665106529096e-05,
+      "loss": 0.6316,
+      "step": 1734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804780729994723e-05,
+      "loss": 0.6733,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68008944403626e-05,
+      "loss": 0.5129,
+      "step": 1736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797006238725315e-05,
+      "loss": 0.6245,
+      "step": 1737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793116126176004e-05,
+      "loss": 0.6113,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789224103808327e-05,
+      "loss": 0.4956,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785330172716495e-05,
+      "loss": 0.6833,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781434333995243e-05,
+      "loss": 0.7236,
+      "step": 1741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777536588739854e-05,
+      "loss": 0.6665,
+      "step": 1742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773636938046136e-05,
+      "loss": 0.6206,
+      "step": 1743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676973538301044e-05,
+      "loss": 0.6411,
+      "step": 1744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765831924729646e-05,
+      "loss": 0.6199,
+      "step": 1745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676192656430118e-05,
+      "loss": 0.6052,
+      "step": 1746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758019302822993e-05,
+      "loss": 0.5569,
+      "step": 1747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6754110141393574e-05,
+      "loss": 0.5999,
+      "step": 1748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750199081111944e-05,
+      "loss": 0.7063,
+      "step": 1749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674628612307766e-05,
+      "loss": 0.5073,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742371268390814e-05,
+      "loss": 0.6243,
+      "step": 1751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738454518152026e-05,
+      "loss": 0.4587,
+      "step": 1752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734535873462455e-05,
+      "loss": 0.515,
+      "step": 1753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730615335423786e-05,
+      "loss": 0.604,
+      "step": 1754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726692905138247e-05,
+      "loss": 0.543,
+      "step": 1755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722768583708582e-05,
+      "loss": 0.5601,
+      "step": 1756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718842372238083e-05,
+      "loss": 0.5451,
+      "step": 1757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714914271830562e-05,
+      "loss": 0.5684,
+      "step": 1758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671098428359037e-05,
+      "loss": 0.5652,
+      "step": 1759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707052408622382e-05,
+      "loss": 0.5742,
+      "step": 1760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703118648032004e-05,
+      "loss": 0.552,
+      "step": 1761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669918300292518e-05,
+      "loss": 0.6394,
+      "step": 1762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695245474408373e-05,
+      "loss": 0.4858,
+      "step": 1763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.5181,
+      "step": 1764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687364771573337e-05,
+      "loss": 0.5742,
+      "step": 1765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683421599470697e-05,
+      "loss": 0.5923,
+      "step": 1766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679476548389235e-05,
+      "loss": 0.6143,
+      "step": 1767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667552961943807e-05,
+      "loss": 0.6045,
+      "step": 1768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671580813726843e-05,
+      "loss": 0.5735,
+      "step": 1769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667630132365725e-05,
+      "loss": 0.4619,
+      "step": 1770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663677576465404e-05,
+      "loss": 0.6177,
+      "step": 1771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659723147137113e-05,
+      "loss": 0.6494,
+      "step": 1772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665576684549259e-05,
+      "loss": 0.5964,
+      "step": 1773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651808672644114e-05,
+      "loss": 0.7864,
+      "step": 1774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664784862970449e-05,
+      "loss": 0.6685,
+      "step": 1775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664388671778705e-05,
+      "loss": 0.5015,
+      "step": 1776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639922938005634e-05,
+      "loss": 0.6514,
+      "step": 1777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635957291474625e-05,
+      "loss": 0.5925,
+      "step": 1778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631989779308932e-05,
+      "loss": 0.4983,
+      "step": 1779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662802040262398e-05,
+      "loss": 0.5835,
+      "step": 1780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624049162535715e-05,
+      "loss": 0.5055,
+      "step": 1781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620076060160618e-05,
+      "loss": 0.5286,
+      "step": 1782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616101096615686e-05,
+      "loss": 0.5837,
+      "step": 1783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661212427301844e-05,
+      "loss": 0.6401,
+      "step": 1784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6608145590486926e-05,
+      "loss": 0.6068,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604165050139712e-05,
+      "loss": 0.6611,
+      "step": 1786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600182653095892e-05,
+      "loss": 0.7046,
+      "step": 1787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659619840047507e-05,
+      "loss": 0.6467,
+      "step": 1788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592212293397386e-05,
+      "loss": 0.4749,
+      "step": 1789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588224332983492e-05,
+      "loss": 0.6719,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584234520354568e-05,
+      "loss": 0.6091,
+      "step": 1791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580242856632306e-05,
+      "loss": 0.5674,
+      "step": 1792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576249342938927e-05,
+      "loss": 0.6143,
+      "step": 1793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657225398039717e-05,
+      "loss": 0.6042,
+      "step": 1794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568256770130285e-05,
+      "loss": 0.5535,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564257713262056e-05,
+      "loss": 0.6316,
+      "step": 1796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560256810916775e-05,
+      "loss": 0.6816,
+      "step": 1797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556254064219257e-05,
+      "loss": 0.6589,
+      "step": 1798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552249474294836e-05,
+      "loss": 0.4781,
+      "step": 1799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548243042269366e-05,
+      "loss": 0.6516,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654423476926921e-05,
+      "loss": 0.5508,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654022465642126e-05,
+      "loss": 0.6172,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653621270485292e-05,
+      "loss": 0.6652,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532198915692105e-05,
+      "loss": 0.5542,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528183290067262e-05,
+      "loss": 0.6787,
+      "step": 1805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524165829107336e-05,
+      "loss": 0.5872,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520146533941805e-05,
+      "loss": 0.6536,
+      "step": 1807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516125405700655e-05,
+      "loss": 0.5092,
+      "step": 1808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512102445514376e-05,
+      "loss": 0.4819,
+      "step": 1809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508077654513992e-05,
+      "loss": 0.4717,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650405103383103e-05,
+      "loss": 0.6787,
+      "step": 1811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500022584597544e-05,
+      "loss": 0.7744,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495992307946086e-05,
+      "loss": 0.5162,
+      "step": 1813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649196020500973e-05,
+      "loss": 0.5986,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648792627692206e-05,
+      "loss": 0.6709,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483890524817178e-05,
+      "loss": 0.4917,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64798529498297e-05,
+      "loss": 0.6516,
+      "step": 1817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475813553094742e-05,
+      "loss": 0.5637,
+      "step": 1818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647177233574795e-05,
+      "loss": 0.6858,
+      "step": 1819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646772929892547e-05,
+      "loss": 0.6294,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463684443763955e-05,
+      "loss": 0.6267,
+      "step": 1821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645963777140059e-05,
+      "loss": 0.6492,
+      "step": 1822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645558928297305e-05,
+      "loss": 0.5251,
+      "step": 1823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451538979619527e-05,
+      "loss": 0.5706,
+      "step": 1824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644748686247872e-05,
+      "loss": 0.6838,
+      "step": 1825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644343293268986e-05,
+      "loss": 0.595,
+      "step": 1826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643937719139265e-05,
+      "loss": 0.5522,
+      "step": 1827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435319639727333e-05,
+      "loss": 0.6033,
+      "step": 1828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431260278834644e-05,
+      "loss": 0.5667,
+      "step": 1829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642719910985584e-05,
+      "loss": 0.655,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642313613393267e-05,
+      "loss": 0.6797,
+      "step": 1831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419071352207413e-05,
+      "loss": 0.5752,
+      "step": 1832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415004765822825e-05,
+      "loss": 0.6599,
+      "step": 1833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64109363759222e-05,
+      "loss": 0.6272,
+      "step": 1834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406866183649327e-05,
+      "loss": 0.6465,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402794190148497e-05,
+      "loss": 0.7314,
+      "step": 1836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398720396564506e-05,
+      "loss": 0.5916,
+      "step": 1837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639464480404267e-05,
+      "loss": 0.6733,
+      "step": 1838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390567413728796e-05,
+      "loss": 0.5868,
+      "step": 1839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638648822676921e-05,
+      "loss": 0.5986,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382407244310726e-05,
+      "loss": 0.6145,
+      "step": 1841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378324467500676e-05,
+      "loss": 0.6138,
+      "step": 1842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.5808,
+      "step": 1843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370153535417725e-05,
+      "loss": 0.6382,
+      "step": 1844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366065382441996e-05,
+      "loss": 0.5459,
+      "step": 1845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636197543970906e-05,
+      "loss": 0.5632,
+      "step": 1846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635788370836876e-05,
+      "loss": 0.6167,
+      "step": 1847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353790189571447e-05,
+      "loss": 0.6282,
+      "step": 1848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349694884467976e-05,
+      "loss": 0.5483,
+      "step": 1849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345597794209703e-05,
+      "loss": 0.6628,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634149891994848e-05,
+      "loss": 0.7393,
+      "step": 1851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633739826283667e-05,
+      "loss": 0.6814,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633329582402713e-05,
+      "loss": 0.5369,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329191604673228e-05,
+      "loss": 0.626,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325085605928813e-05,
+      "loss": 0.6091,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320977828948257e-05,
+      "loss": 0.498,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316868274886415e-05,
+      "loss": 0.5833,
+      "step": 1857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312756944898653e-05,
+      "loss": 0.5991,
+      "step": 1858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630864384014083e-05,
+      "loss": 0.5815,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304528961769295e-05,
+      "loss": 0.5012,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300412310940923e-05,
+      "loss": 0.614,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629629388881306e-05,
+      "loss": 0.5204,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292173696543564e-05,
+      "loss": 0.6558,
+      "step": 1863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288051735290782e-05,
+      "loss": 0.422,
+      "step": 1864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283928006213568e-05,
+      "loss": 0.6311,
+      "step": 1865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627980251047127e-05,
+      "loss": 0.5967,
+      "step": 1866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275675249223723e-05,
+      "loss": 0.5835,
+      "step": 1867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271546223631275e-05,
+      "loss": 0.5686,
+      "step": 1868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267415434854756e-05,
+      "loss": 0.7627,
+      "step": 1869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263282884055495e-05,
+      "loss": 0.6638,
+      "step": 1870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625914857239532e-05,
+      "loss": 0.5913,
+      "step": 1871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625501250103656e-05,
+      "loss": 0.6346,
+      "step": 1872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250874671142022e-05,
+      "loss": 0.5513,
+      "step": 1873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624673508387502e-05,
+      "loss": 0.6289,
+      "step": 1874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624259374039936e-05,
+      "loss": 0.6204,
+      "step": 1875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6238450641879335e-05,
+      "loss": 0.4846,
+      "step": 1876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234305789479744e-05,
+      "loss": 0.6226,
+      "step": 1877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230159184365866e-05,
+      "loss": 0.5596,
+      "step": 1878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226010827703483e-05,
+      "loss": 0.51,
+      "step": 1879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221860720658862e-05,
+      "loss": 0.7498,
+      "step": 1880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217708864398763e-05,
+      "loss": 0.592,
+      "step": 1881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621355526009045e-05,
+      "loss": 0.6279,
+      "step": 1882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620939990890165e-05,
+      "loss": 0.5276,
+      "step": 1883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620524281200062e-05,
+      "loss": 0.6196,
+      "step": 1884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201083970556074e-05,
+      "loss": 0.5564,
+      "step": 1885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196923385737232e-05,
+      "loss": 0.6853,
+      "step": 1886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619276105871381e-05,
+      "loss": 0.4856,
+      "step": 1887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618859699065599e-05,
+      "loss": 0.5222,
+      "step": 1888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184431182734475e-05,
+      "loss": 0.6382,
+      "step": 1889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180263636120433e-05,
+      "loss": 0.5491,
+      "step": 1890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176094351985527e-05,
+      "loss": 0.5864,
+      "step": 1891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171923331501922e-05,
+      "loss": 0.6326,
+      "step": 1892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616775057584225e-05,
+      "loss": 0.5293,
+      "step": 1893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163576086179636e-05,
+      "loss": 0.595,
+      "step": 1894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615939986368771e-05,
+      "loss": 0.6565,
+      "step": 1895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155221909540565e-05,
+      "loss": 0.4819,
+      "step": 1896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151042224912797e-05,
+      "loss": 0.6057,
+      "step": 1897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146860810979488e-05,
+      "loss": 0.6445,
+      "step": 1898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142677668916194e-05,
+      "loss": 0.6846,
+      "step": 1899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613849279989897e-05,
+      "loss": 0.5713,
+      "step": 1900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134306205104345e-05,
+      "loss": 0.7002,
+      "step": 1901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130117885709346e-05,
+      "loss": 0.5905,
+      "step": 1902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125927842891476e-05,
+      "loss": 0.5408,
+      "step": 1903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121736077828725e-05,
+      "loss": 0.6403,
+      "step": 1904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117542591699562e-05,
+      "loss": 0.5862,
+      "step": 1905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113347385682953e-05,
+      "loss": 0.5658,
+      "step": 1906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109150460958333e-05,
+      "loss": 0.5699,
+      "step": 1907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104951818705626e-05,
+      "loss": 0.7712,
+      "step": 1908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100751460105244e-05,
+      "loss": 0.4988,
+      "step": 1909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096549386338075e-05,
+      "loss": 0.7024,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092345598585493e-05,
+      "loss": 0.6597,
+      "step": 1911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088140098029344e-05,
+      "loss": 0.7339,
+      "step": 1912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083932885851974e-05,
+      "loss": 0.7303,
+      "step": 1913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079723963236192e-05,
+      "loss": 0.5889,
+      "step": 1914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60755133313653e-05,
+      "loss": 0.6189,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071300991423076e-05,
+      "loss": 0.6316,
+      "step": 1916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067086944593774e-05,
+      "loss": 0.5459,
+      "step": 1917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062871192062135e-05,
+      "loss": 0.6313,
+      "step": 1918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058653735013376e-05,
+      "loss": 0.5439,
+      "step": 1919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054434574633195e-05,
+      "loss": 0.5886,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050213712107768e-05,
+      "loss": 0.6504,
+      "step": 1921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 0.5315,
+      "step": 1922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604176688536827e-05,
+      "loss": 0.6106,
+      "step": 1923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037540923528943e-05,
+      "loss": 0.6147,
+      "step": 1924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033313264293853e-05,
+      "loss": 0.5222,
+      "step": 1925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029083908851565e-05,
+      "loss": 0.5002,
+      "step": 1926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024852858391125e-05,
+      "loss": 0.5051,
+      "step": 1927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602062011410205e-05,
+      "loss": 0.6521,
+      "step": 1928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016385677174334e-05,
+      "loss": 0.6206,
+      "step": 1929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601214954879845e-05,
+      "loss": 0.5413,
+      "step": 1930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600791173016534e-05,
+      "loss": 0.5791,
+      "step": 1931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003672222466428e-05,
+      "loss": 0.6433,
+      "step": 1932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999431026893607e-05,
+      "loss": 0.5481,
+      "step": 1933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599518814463925e-05,
+      "loss": 0.5264,
+      "step": 1934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990943576896208e-05,
+      "loss": 0.6509,
+      "step": 1935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598669732485779e-05,
+      "loss": 0.625,
+      "step": 1936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982449389717794e-05,
+      "loss": 0.6641,
+      "step": 1937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597819977267048e-05,
+      "loss": 0.5249,
+      "step": 1938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973948474910595e-05,
+      "loss": 0.7517,
+      "step": 1939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969695497633348e-05,
+      "loss": 0.7017,
+      "step": 1940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596544084203442e-05,
+      "loss": 0.5776,
+      "step": 1941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596118450930996e-05,
+      "loss": 0.8159,
+      "step": 1942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956926500656602e-05,
+      "loss": 0.5977,
+      "step": 1943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952666817271443e-05,
+      "loss": 0.5874,
+      "step": 1944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948405460352048e-05,
+      "loss": 0.595,
+      "step": 1945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594414243109646e-05,
+      "loss": 0.6333,
+      "step": 1946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939877730703184e-05,
+      "loss": 0.5308,
+      "step": 1947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59356113603712e-05,
+      "loss": 0.5627,
+      "step": 1948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593134332129996e-05,
+      "loss": 0.5146,
+      "step": 1949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592707361468938e-05,
+      "loss": 0.6118,
+      "step": 1950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922802241739844e-05,
+      "loss": 0.5386,
+      "step": 1951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591852920365221e-05,
+      "loss": 0.5983,
+      "step": 1952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914254501627797e-05,
+      "loss": 0.5378,
+      "step": 1953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909978136868396e-05,
+      "loss": 0.5212,
+      "step": 1954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590570011057627e-05,
+      "loss": 0.5574,
+      "step": 1955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590142042395414e-05,
+      "loss": 0.6331,
+      "step": 1956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897139078205204e-05,
+      "loss": 0.5959,
+      "step": 1957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892856074533114e-05,
+      "loss": 0.6162,
+      "step": 1958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888571414141997e-05,
+      "loss": 0.6438,
+      "step": 1959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884285098236444e-05,
+      "loss": 0.5298,
+      "step": 1960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587999712802151e-05,
+      "loss": 0.5637,
+      "step": 1961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587570750470272e-05,
+      "loss": 0.6455,
+      "step": 1962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587141622948605e-05,
+      "loss": 0.5229,
+      "step": 1963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586712330357796e-05,
+      "loss": 0.6726,
+      "step": 1964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586282872818536e-05,
+      "loss": 0.6746,
+      "step": 1965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585853250451563e-05,
+      "loss": 0.5002,
+      "step": 1966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854234633776612e-05,
+      "loss": 0.5308,
+      "step": 1967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849935117176608e-05,
+      "loss": 0.6343,
+      "step": 1968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845633955924382e-05,
+      "loss": 0.6826,
+      "step": 1969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841331151229173e-05,
+      "loss": 0.6536,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583702670430066e-05,
+      "loss": 0.6321,
+      "step": 1971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832720616349006e-05,
+      "loss": 0.615,
+      "step": 1972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828412888584825e-05,
+      "loss": 0.7319,
+      "step": 1973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824103522219188e-05,
+      "loss": 0.7998,
+      "step": 1974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819792518463634e-05,
+      "loss": 0.428,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815479878530166e-05,
+      "loss": 0.5188,
+      "step": 1976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581116560363123e-05,
+      "loss": 0.5886,
+      "step": 1977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806849694979742e-05,
+      "loss": 0.6206,
+      "step": 1978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802532153789085e-05,
+      "loss": 0.6558,
+      "step": 1979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798212981273094e-05,
+      "loss": 0.5339,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793892178646055e-05,
+      "loss": 0.7476,
+      "step": 1981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789569747122725e-05,
+      "loss": 0.4429,
+      "step": 1982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785245687918312e-05,
+      "loss": 0.5916,
+      "step": 1983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780920002248484e-05,
+      "loss": 0.5381,
+      "step": 1984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776592691329368e-05,
+      "loss": 0.5818,
+      "step": 1985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772263756377535e-05,
+      "loss": 0.668,
+      "step": 1986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576793319861003e-05,
+      "loss": 0.5964,
+      "step": 1987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576360101924435e-05,
+      "loss": 0.5535,
+      "step": 1988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759267219498444e-05,
+      "loss": 0.594,
+      "step": 1989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754931800590712e-05,
+      "loss": 0.616,
+      "step": 1990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750594763740018e-05,
+      "loss": 0.6523,
+      "step": 1991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746256110165673e-05,
+      "loss": 0.5481,
+      "step": 1992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5741915841087457e-05,
+      "loss": 0.6117,
+      "step": 1993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573757395772559e-05,
+      "loss": 0.7393,
+      "step": 1994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573323046130075e-05,
+      "loss": 0.5645,
+      "step": 1995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5728885353034063e-05,
+      "loss": 0.551,
+      "step": 1996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724538634147122e-05,
+      "loss": 0.5806,
+      "step": 1997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572019030586196e-05,
+      "loss": 0.4331,
+      "step": 1998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571584036940107e-05,
+      "loss": 0.4194,
+      "step": 1999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571148882598739e-05,
+      "loss": 0.5967,
+      "step": 2000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 0.5737,
+      "step": 2001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57027809231957e-05,
+      "loss": 0.7148,
+      "step": 2002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569842456626583e-05,
+      "loss": 0.6135,
+      "step": 2003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569406660727945e-05,
+      "loss": 0.6411,
+      "step": 2004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689707047461772e-05,
+      "loss": 0.7783,
+      "step": 2005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685345888038432e-05,
+      "loss": 0.708,
+      "step": 2006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568098313023553e-05,
+      "loss": 0.6111,
+      "step": 2007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676618775279615e-05,
+      "loss": 0.8076,
+      "step": 2008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672252824397683e-05,
+      "loss": 0.5564,
+      "step": 2009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667885278817173e-05,
+      "loss": 0.7134,
+      "step": 2010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663516139765983e-05,
+      "loss": 0.615,
+      "step": 2011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659145408472454e-05,
+      "loss": 0.5083,
+      "step": 2012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654773086165373e-05,
+      "loss": 0.6245,
+      "step": 2013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650399174073977e-05,
+      "loss": 0.479,
+      "step": 2014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646023673427943e-05,
+      "loss": 0.5352,
+      "step": 2015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641646585457408e-05,
+      "loss": 0.5786,
+      "step": 2016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637267911392947e-05,
+      "loss": 0.5906,
+      "step": 2017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632887652465575e-05,
+      "loss": 0.6699,
+      "step": 2018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628505809906763e-05,
+      "loss": 0.5339,
+      "step": 2019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562412238494843e-05,
+      "loss": 0.666,
+      "step": 2020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561973737882292e-05,
+      "loss": 0.6604,
+      "step": 2021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615350792763044e-05,
+      "loss": 0.5371,
+      "step": 2022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610962628002044e-05,
+      "loss": 0.4833,
+      "step": 2023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606572885773613e-05,
+      "loss": 0.7666,
+      "step": 2024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560218156731188e-05,
+      "loss": 0.6365,
+      "step": 2025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597788673851426e-05,
+      "loss": 0.6287,
+      "step": 2026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593394206627266e-05,
+      "loss": 0.6282,
+      "step": 2027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588998166874866e-05,
+      "loss": 0.5957,
+      "step": 2028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584600555830125e-05,
+      "loss": 0.478,
+      "step": 2029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580201374729388e-05,
+      "loss": 0.6384,
+      "step": 2030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557580062480945e-05,
+      "loss": 0.575,
+      "step": 2031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557139830730753e-05,
+      "loss": 0.6162,
+      "step": 2032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55669944234613e-05,
+      "loss": 0.6169,
+      "step": 2033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556258897450887e-05,
+      "loss": 0.5942,
+      "step": 2034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558181961688788e-05,
+      "loss": 0.5786,
+      "step": 2035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553773386240044e-05,
+      "loss": 0.5375,
+      "step": 2036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549363249402064e-05,
+      "loss": 0.6848,
+      "step": 2037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544951552414716e-05,
+      "loss": 0.5837,
+      "step": 2038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540538296518305e-05,
+      "loss": 0.5925,
+      "step": 2039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536123482953577e-05,
+      "loss": 0.5256,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531707112961708e-05,
+      "loss": 0.4913,
+      "step": 2041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527289187784324e-05,
+      "loss": 0.4971,
+      "step": 2042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522869708663477e-05,
+      "loss": 0.6121,
+      "step": 2043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518448676841665e-05,
+      "loss": 0.5525,
+      "step": 2044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551402609356181e-05,
+      "loss": 0.5862,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5509601960067293e-05,
+      "loss": 0.6921,
+      "step": 2046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55051762776019e-05,
+      "loss": 0.5657,
+      "step": 2047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500749047409874e-05,
+      "loss": 0.6069,
+      "step": 2048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549632027073589e-05,
+      "loss": 0.7114,
+      "step": 2049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491889948825054e-05,
+      "loss": 0.6917,
+      "step": 2050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487458082922908e-05,
+      "loss": 0.5847,
+      "step": 2051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548302467427543e-05,
+      "loss": 0.5271,
+      "step": 2052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478589724129027e-05,
+      "loss": 0.6135,
+      "step": 2053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547415323373054e-05,
+      "loss": 0.5898,
+      "step": 2054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469715204327247e-05,
+      "loss": 0.6384,
+      "step": 2055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465275637166857e-05,
+      "loss": 0.5757,
+      "step": 2056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546083453349751e-05,
+      "loss": 0.5767,
+      "step": 2057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456391894567784e-05,
+      "loss": 0.6394,
+      "step": 2058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451947721626676e-05,
+      "loss": 0.6284,
+      "step": 2059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544750201592363e-05,
+      "loss": 0.5632,
+      "step": 2060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443054778708507e-05,
+      "loss": 0.6182,
+      "step": 2061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5438606011231604e-05,
+      "loss": 0.793,
+      "step": 2062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543415571474365e-05,
+      "loss": 0.5667,
+      "step": 2063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54297038904958e-05,
+      "loss": 0.7524,
+      "step": 2064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542525053973965e-05,
+      "loss": 0.5464,
+      "step": 2065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420795663727207e-05,
+      "loss": 0.5286,
+      "step": 2066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416339263710915e-05,
+      "loss": 0.5955,
+      "step": 2067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411881340943655e-05,
+      "loss": 0.5653,
+      "step": 2068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407421896678726e-05,
+      "loss": 0.7405,
+      "step": 2069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402960932169854e-05,
+      "loss": 0.6492,
+      "step": 2070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398498448671197e-05,
+      "loss": 0.5356,
+      "step": 2071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394034447437343e-05,
+      "loss": 0.5771,
+      "step": 2072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389568929723297e-05,
+      "loss": 0.6826,
+      "step": 2073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385101896784497e-05,
+      "loss": 0.4536,
+      "step": 2074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380633349876805e-05,
+      "loss": 0.51,
+      "step": 2075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376163290256516e-05,
+      "loss": 0.6606,
+      "step": 2076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371691719180333e-05,
+      "loss": 0.6411,
+      "step": 2077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367218637905402e-05,
+      "loss": 0.5994,
+      "step": 2078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362744047689285e-05,
+      "loss": 0.55,
+      "step": 2079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.5796,
+      "step": 2080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535379034546586e-05,
+      "loss": 0.6548,
+      "step": 2081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349311235975802e-05,
+      "loss": 0.6677,
+      "step": 2082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534483062257905e-05,
+      "loss": 0.5806,
+      "step": 2083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534034850653528e-05,
+      "loss": 0.7163,
+      "step": 2084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335864889104604e-05,
+      "loss": 0.6104,
+      "step": 2085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331379771547536e-05,
+      "loss": 0.5442,
+      "step": 2086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326893155125033e-05,
+      "loss": 0.5962,
+      "step": 2087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322405041098457e-05,
+      "loss": 0.606,
+      "step": 2088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317915430729602e-05,
+      "loss": 0.6548,
+      "step": 2089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313424325280676e-05,
+      "loss": 0.676,
+      "step": 2090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308931726014307e-05,
+      "loss": 0.5618,
+      "step": 2091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530443763419355e-05,
+      "loss": 0.6956,
+      "step": 2092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299942051081874e-05,
+      "loss": 0.6912,
+      "step": 2093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529544497794317e-05,
+      "loss": 0.4863,
+      "step": 2094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290946416041742e-05,
+      "loss": 0.6201,
+      "step": 2095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5286446366642317e-05,
+      "loss": 0.7891,
+      "step": 2096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528194483101004e-05,
+      "loss": 0.6384,
+      "step": 2097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277441810410478e-05,
+      "loss": 0.5718,
+      "step": 2098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272937306109608e-05,
+      "loss": 0.5468,
+      "step": 2099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268431319373833e-05,
+      "loss": 0.5498,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263923851469957e-05,
+      "loss": 0.5754,
+      "step": 2101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525941490366522e-05,
+      "loss": 0.7075,
+      "step": 2102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254904477227261e-05,
+      "loss": 0.6963,
+      "step": 2103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250392573424152e-05,
+      "loss": 0.6182,
+      "step": 2104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524587919352436e-05,
+      "loss": 0.6055,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241364338796785e-05,
+      "loss": 0.6648,
+      "step": 2106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236848010510733e-05,
+      "loss": 0.4978,
+      "step": 2107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5232330209935922e-05,
+      "loss": 0.5764,
+      "step": 2108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227810938342493e-05,
+      "loss": 0.4078,
+      "step": 2109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223290197000993e-05,
+      "loss": 0.5493,
+      "step": 2110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218767987182384e-05,
+      "loss": 0.5488,
+      "step": 2111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214244310158042e-05,
+      "loss": 0.6843,
+      "step": 2112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520971916719975e-05,
+      "loss": 0.6299,
+      "step": 2113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205192559579716e-05,
+      "loss": 0.47,
+      "step": 2114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200664488570542e-05,
+      "loss": 0.5752,
+      "step": 2115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196134955445261e-05,
+      "loss": 0.7058,
+      "step": 2116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191603961477302e-05,
+      "loss": 0.6426,
+      "step": 2117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187071507940508e-05,
+      "loss": 0.6428,
+      "step": 2118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182537596109139e-05,
+      "loss": 0.5447,
+      "step": 2119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178002227257855e-05,
+      "loss": 0.7178,
+      "step": 2120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173465402661735e-05,
+      "loss": 0.6626,
+      "step": 2121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516892712359626e-05,
+      "loss": 0.6089,
+      "step": 2122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164387391337325e-05,
+      "loss": 0.5525,
+      "step": 2123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515984620716123e-05,
+      "loss": 0.6462,
+      "step": 2124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155303572344686e-05,
+      "loss": 0.6223,
+      "step": 2125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515075948816481e-05,
+      "loss": 0.6213,
+      "step": 2126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514621395589912e-05,
+      "loss": 0.6626,
+      "step": 2127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141666976825562e-05,
+      "loss": 0.4827,
+      "step": 2128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137118552222468e-05,
+      "loss": 0.6704,
+      "step": 2129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132568683368578e-05,
+      "loss": 0.5576,
+      "step": 2130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128017371543052e-05,
+      "loss": 0.7637,
+      "step": 2131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512346461802544e-05,
+      "loss": 0.6885,
+      "step": 2132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511891042409571e-05,
+      "loss": 0.6133,
+      "step": 2133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114354791034225e-05,
+      "loss": 0.6335,
+      "step": 2134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109797720121758e-05,
+      "loss": 0.6323,
+      "step": 2135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105239212639484e-05,
+      "loss": 0.7207,
+      "step": 2136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100679269868988e-05,
+      "loss": 0.5962,
+      "step": 2137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096117893092245e-05,
+      "loss": 0.6819,
+      "step": 2138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091555083591647e-05,
+      "loss": 0.5564,
+      "step": 2139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086990842649984e-05,
+      "loss": 0.3983,
+      "step": 2140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082425171550448e-05,
+      "loss": 0.5509,
+      "step": 2141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507785807157663e-05,
+      "loss": 0.4998,
+      "step": 2142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073289544012526e-05,
+      "loss": 0.5806,
+      "step": 2143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068719590142537e-05,
+      "loss": 0.6472,
+      "step": 2144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064148211251457e-05,
+      "loss": 0.5414,
+      "step": 2145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059575408624483e-05,
+      "loss": 0.6792,
+      "step": 2146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055001183547218e-05,
+      "loss": 0.6846,
+      "step": 2147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505042553730566e-05,
+      "loss": 0.7026,
+      "step": 2148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045848471186206e-05,
+      "loss": 0.6497,
+      "step": 2149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041269986475654e-05,
+      "loss": 0.5977,
+      "step": 2150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036690084461201e-05,
+      "loss": 0.624,
+      "step": 2151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503210876643044e-05,
+      "loss": 0.6292,
+      "step": 2152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027526033671365e-05,
+      "loss": 0.6938,
+      "step": 2153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022941887472369e-05,
+      "loss": 0.6062,
+      "step": 2154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018356329122237e-05,
+      "loss": 0.5653,
+      "step": 2155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013769359910156e-05,
+      "loss": 0.5378,
+      "step": 2156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009180981125702e-05,
+      "loss": 0.6384,
+      "step": 2157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004591194058859e-05,
+      "loss": 0.6865,
+      "step": 2158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7178,
+      "step": 2159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995407400239896e-05,
+      "loss": 0.5889,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990813396069707e-05,
+      "loss": 0.616,
+      "step": 2161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986217988780993e-05,
+      "loss": 0.6421,
+      "step": 2162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981621179665713e-05,
+      "loss": 0.6226,
+      "step": 2163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977022970016211e-05,
+      "loss": 0.5527,
+      "step": 2164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497242336112523e-05,
+      "loss": 0.655,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967822354285904e-05,
+      "loss": 0.644,
+      "step": 2166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963219950791762e-05,
+      "loss": 0.6089,
+      "step": 2167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958616151936726e-05,
+      "loss": 0.7461,
+      "step": 2168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4954010959015106e-05,
+      "loss": 0.781,
+      "step": 2169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949404373321612e-05,
+      "loss": 0.6213,
+      "step": 2170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944796396151338e-05,
+      "loss": 0.6016,
+      "step": 2171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940187028799768e-05,
+      "loss": 0.5393,
+      "step": 2172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935576272562786e-05,
+      "loss": 0.6521,
+      "step": 2173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493096412873666e-05,
+      "loss": 0.6758,
+      "step": 2174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926350598618048e-05,
+      "loss": 0.6311,
+      "step": 2175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921735683504001e-05,
+      "loss": 0.5874,
+      "step": 2176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917119384691956e-05,
+      "loss": 0.4536,
+      "step": 2177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491250170347974e-05,
+      "loss": 0.6558,
+      "step": 2178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490788264116557e-05,
+      "loss": 0.5823,
+      "step": 2179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490326219904805e-05,
+      "loss": 0.553,
+      "step": 2180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489864037842617e-05,
+      "loss": 0.6028,
+      "step": 2181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894017180599317e-05,
+      "loss": 0.4629,
+      "step": 2182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488939260686725e-05,
+      "loss": 0.7578,
+      "step": 2183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884766658530126e-05,
+      "loss": 0.5742,
+      "step": 2184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880139336888485e-05,
+      "loss": 0.6162,
+      "step": 2185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487551064324325e-05,
+      "loss": 0.6792,
+      "step": 2186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487088057889574e-05,
+      "loss": 0.6028,
+      "step": 2187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866249145147648e-05,
+      "loss": 0.6526,
+      "step": 2188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486161634330106e-05,
+      "loss": 0.707,
+      "step": 2189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856982174658439e-05,
+      "loss": 0.6428,
+      "step": 2190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852346640522636e-05,
+      "loss": 0.5148,
+      "step": 2191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847709742196892e-05,
+      "loss": 0.6848,
+      "step": 2192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843071480984816e-05,
+      "loss": 0.698,
+      "step": 2193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838431858190423e-05,
+      "loss": 0.4423,
+      "step": 2194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833790875118086e-05,
+      "loss": 0.6218,
+      "step": 2195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829148533072577e-05,
+      "loss": 0.5266,
+      "step": 2196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824504833359043e-05,
+      "loss": 0.7083,
+      "step": 2197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481985977728302e-05,
+      "loss": 0.6804,
+      "step": 2198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815213366150414e-05,
+      "loss": 0.6306,
+      "step": 2199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4810565601267521e-05,
+      "loss": 0.6729,
+      "step": 2200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805916483941016e-05,
+      "loss": 0.6433,
+      "step": 2201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801266015477954e-05,
+      "loss": 0.5371,
+      "step": 2202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796614197185763e-05,
+      "loss": 0.5918,
+      "step": 2203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791961030372264e-05,
+      "loss": 0.8091,
+      "step": 2204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787306516345638e-05,
+      "loss": 0.6611,
+      "step": 2205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782650656414472e-05,
+      "loss": 0.6477,
+      "step": 2206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777993451887707e-05,
+      "loss": 0.6624,
+      "step": 2207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773334904074668e-05,
+      "loss": 0.5095,
+      "step": 2208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768675014285063e-05,
+      "loss": 0.5103,
+      "step": 2209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764013783828974e-05,
+      "loss": 0.666,
+      "step": 2210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759351214016864e-05,
+      "loss": 0.6399,
+      "step": 2211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754687306159565e-05,
+      "loss": 0.5494,
+      "step": 2212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475002206156829e-05,
+      "loss": 0.6606,
+      "step": 2213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745355481554629e-05,
+      "loss": 0.4753,
+      "step": 2214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740687567430545e-05,
+      "loss": 0.6514,
+      "step": 2215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736018320508373e-05,
+      "loss": 0.5452,
+      "step": 2216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473134774210083e-05,
+      "loss": 0.6621,
+      "step": 2217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726675833521004e-05,
+      "loss": 0.7483,
+      "step": 2218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722002596082357e-05,
+      "loss": 0.6609,
+      "step": 2219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717328031098719e-05,
+      "loss": 0.4734,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47126521398843e-05,
+      "loss": 0.71,
+      "step": 2221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707974923753683e-05,
+      "loss": 0.5466,
+      "step": 2222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703296384021824e-05,
+      "loss": 0.5857,
+      "step": 2223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698616522004046e-05,
+      "loss": 0.5842,
+      "step": 2224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693935339016046e-05,
+      "loss": 0.7395,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4689252836373895e-05,
+      "loss": 0.6379,
+      "step": 2226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468456901539403e-05,
+      "loss": 0.6892,
+      "step": 2227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679883877393264e-05,
+      "loss": 0.5347,
+      "step": 2228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675197423688779e-05,
+      "loss": 0.5513,
+      "step": 2229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670509655598123e-05,
+      "loss": 0.5222,
+      "step": 2230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665820574439218e-05,
+      "loss": 0.5044,
+      "step": 2231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661130181530351e-05,
+      "loss": 0.7012,
+      "step": 2232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656438478190185e-05,
+      "loss": 0.4189,
+      "step": 2233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465174546573774e-05,
+      "loss": 0.5701,
+      "step": 2234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647051145492415e-05,
+      "loss": 0.5208,
+      "step": 2235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642355518773973e-05,
+      "loss": 0.6072,
+      "step": 2236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637658586902542e-05,
+      "loss": 0.5754,
+      "step": 2237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 0.6047,
+      "step": 2238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628260812983066e-05,
+      "loss": 0.541,
+      "step": 2239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623559973577113e-05,
+      "loss": 0.5627,
+      "step": 2240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618857834302358e-05,
+      "loss": 0.6196,
+      "step": 2241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614154396480761e-05,
+      "loss": 0.5852,
+      "step": 2242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609449661434646e-05,
+      "loss": 0.782,
+      "step": 2243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46047436304867e-05,
+      "loss": 0.5469,
+      "step": 2244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600036304959978e-05,
+      "loss": 0.6804,
+      "step": 2245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595327686177907e-05,
+      "loss": 0.7441,
+      "step": 2246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590617775464262e-05,
+      "loss": 0.6367,
+      "step": 2247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458590657414319e-05,
+      "loss": 0.5706,
+      "step": 2248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581194083539201e-05,
+      "loss": 0.5073,
+      "step": 2249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576480304977166e-05,
+      "loss": 0.6077,
+      "step": 2250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571765239782314e-05,
+      "loss": 0.5969,
+      "step": 2251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456704888928024e-05,
+      "loss": 0.5259,
+      "step": 2252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562331254796907e-05,
+      "loss": 0.4924,
+      "step": 2253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557612337658626e-05,
+      "loss": 0.6565,
+      "step": 2254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552892139192074e-05,
+      "loss": 0.6506,
+      "step": 2255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454817066072429e-05,
+      "loss": 0.5635,
+      "step": 2256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454344790358267e-05,
+      "loss": 0.626,
+      "step": 2257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538723869094976e-05,
+      "loss": 0.499,
+      "step": 2258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533998558589319e-05,
+      "loss": 0.6467,
+      "step": 2259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529271973394178e-05,
+      "loss": 0.5125,
+      "step": 2260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524544114838386e-05,
+      "loss": 0.6345,
+      "step": 2261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451981498425113e-05,
+      "loss": 0.4689,
+      "step": 2262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515084582961963e-05,
+      "loss": 0.5493,
+      "step": 2263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451035291230079e-05,
+      "loss": 0.7229,
+      "step": 2264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505619973597873e-05,
+      "loss": 0.5662,
+      "step": 2265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500885768183834e-05,
+      "loss": 0.5059,
+      "step": 2266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496150297389648e-05,
+      "loss": 0.5474,
+      "step": 2267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491413562546646e-05,
+      "loss": 0.7075,
+      "step": 2268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448667556498651e-05,
+      "loss": 0.5586,
+      "step": 2269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448193630604129e-05,
+      "loss": 0.6729,
+      "step": 2270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447719578704338e-05,
+      "loss": 0.4662,
+      "step": 2271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472454009325528e-05,
+      "loss": 0.5742,
+      "step": 2272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467710974220839e-05,
+      "loss": 0.6069,
+      "step": 2273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462966683062773e-05,
+      "loss": 0.5959,
+      "step": 2274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458221137185142e-05,
+      "loss": 0.592,
+      "step": 2275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453474337922107e-05,
+      "loss": 0.5752,
+      "step": 2276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448726286608186e-05,
+      "loss": 0.5767,
+      "step": 2277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444397698457825e-05,
+      "loss": 0.6409,
+      "step": 2278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439226433167513e-05,
+      "loss": 0.6299,
+      "step": 2279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434474633711548e-05,
+      "loss": 0.6118,
+      "step": 2280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429721587546277e-05,
+      "loss": 0.6042,
+      "step": 2281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424967296007977e-05,
+      "loss": 0.636,
+      "step": 2282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420211760433267e-05,
+      "loss": 0.5481,
+      "step": 2283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415454982159121e-05,
+      "loss": 0.7466,
+      "step": 2284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410696962522859e-05,
+      "loss": 0.5442,
+      "step": 2285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405937702862151e-05,
+      "loss": 0.5593,
+      "step": 2286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440117720451502e-05,
+      "loss": 0.6392,
+      "step": 2287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396415468819828e-05,
+      "loss": 0.7139,
+      "step": 2288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4391652497115299e-05,
+      "loss": 0.6843,
+      "step": 2289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386888290740493e-05,
+      "loss": 0.6843,
+      "step": 2290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382122851034815e-05,
+      "loss": 0.7485,
+      "step": 2291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377356179338027e-05,
+      "loss": 0.5427,
+      "step": 2292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372588276990227e-05,
+      "loss": 0.5281,
+      "step": 2293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367819145331872e-05,
+      "loss": 0.5493,
+      "step": 2294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363048785703753e-05,
+      "loss": 0.7021,
+      "step": 2295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358277199447007e-05,
+      "loss": 0.6802,
+      "step": 2296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353504387903126e-05,
+      "loss": 0.5662,
+      "step": 2297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348730352413932e-05,
+      "loss": 0.6855,
+      "step": 2298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343955094321605e-05,
+      "loss": 0.5732,
+      "step": 2299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339178614968654e-05,
+      "loss": 0.6689,
+      "step": 2300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334400915697949e-05,
+      "loss": 0.5295,
+      "step": 2301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432962199785269e-05,
+      "loss": 0.6477,
+      "step": 2302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432484186277642e-05,
+      "loss": 0.5745,
+      "step": 2303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320060511813032e-05,
+      "loss": 0.7061,
+      "step": 2304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431527794630675e-05,
+      "loss": 0.658,
+      "step": 2305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310494167602152e-05,
+      "loss": 0.6577,
+      "step": 2306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305709177044149e-05,
+      "loss": 0.5911,
+      "step": 2307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300922975977995e-05,
+      "loss": 0.6023,
+      "step": 2308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429613556574928e-05,
+      "loss": 0.7104,
+      "step": 2309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291346947703944e-05,
+      "loss": 0.7676,
+      "step": 2310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286557123188254e-05,
+      "loss": 0.6091,
+      "step": 2311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281766093548824e-05,
+      "loss": 0.6604,
+      "step": 2312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276973860132605e-05,
+      "loss": 0.5354,
+      "step": 2313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272180424286895e-05,
+      "loss": 0.5969,
+      "step": 2314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426738578735931e-05,
+      "loss": 0.4861,
+      "step": 2315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262589950697821e-05,
+      "loss": 0.511,
+      "step": 2316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.5598,
+      "step": 2317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252994683566674e-05,
+      "loss": 0.5933,
+      "step": 2318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248195255794634e-05,
+      "loss": 0.566,
+      "step": 2319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424339463368392e-05,
+      "loss": 0.6541,
+      "step": 2320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238592818584181e-05,
+      "loss": 0.5212,
+      "step": 2321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233789811845397e-05,
+      "loss": 0.4719,
+      "step": 2322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422898561481789e-05,
+      "loss": 0.6438,
+      "step": 2323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224180228852314e-05,
+      "loss": 0.4843,
+      "step": 2324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219373655299653e-05,
+      "loss": 0.6367,
+      "step": 2325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214565895511233e-05,
+      "loss": 0.5603,
+      "step": 2326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209756950838704e-05,
+      "loss": 0.5842,
+      "step": 2327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204946822634056e-05,
+      "loss": 0.51,
+      "step": 2328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200135512249606e-05,
+      "loss": 0.4221,
+      "step": 2329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195323021038014e-05,
+      "loss": 0.5452,
+      "step": 2330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190509350352262e-05,
+      "loss": 0.5513,
+      "step": 2331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185694501545664e-05,
+      "loss": 0.5642,
+      "step": 2332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4180878475971869e-05,
+      "loss": 0.5879,
+      "step": 2333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176061274984858e-05,
+      "loss": 0.5453,
+      "step": 2334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171242899938936e-05,
+      "loss": 0.5193,
+      "step": 2335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166423352188746e-05,
+      "loss": 0.5493,
+      "step": 2336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161602633089256e-05,
+      "loss": 0.7932,
+      "step": 2337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156780743995763e-05,
+      "loss": 0.5879,
+      "step": 2338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151957686263895e-05,
+      "loss": 0.4885,
+      "step": 2339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147133461249609e-05,
+      "loss": 0.5049,
+      "step": 2340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142308070309183e-05,
+      "loss": 0.5938,
+      "step": 2341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137481514799237e-05,
+      "loss": 0.6499,
+      "step": 2342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132653796076703e-05,
+      "loss": 0.6221,
+      "step": 2343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127824915498852e-05,
+      "loss": 0.5527,
+      "step": 2344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122994874423272e-05,
+      "loss": 0.646,
+      "step": 2345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118163674207886e-05,
+      "loss": 0.8374,
+      "step": 2346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113331316210937e-05,
+      "loss": 0.5592,
+      "step": 2347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108497801790995e-05,
+      "loss": 0.6895,
+      "step": 2348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103663132306957e-05,
+      "loss": 0.584,
+      "step": 2349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098827309118042e-05,
+      "loss": 0.5935,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093990333583798e-05,
+      "loss": 0.6973,
+      "step": 2351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4089152207064084e-05,
+      "loss": 0.6211,
+      "step": 2352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084312930919102e-05,
+      "loss": 0.5018,
+      "step": 2353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079472506509365e-05,
+      "loss": 0.5271,
+      "step": 2354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074630935195707e-05,
+      "loss": 0.6621,
+      "step": 2355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069788218339296e-05,
+      "loss": 0.6204,
+      "step": 2356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4064944357301604e-05,
+      "loss": 0.5605,
+      "step": 2357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060099353444445e-05,
+      "loss": 0.4829,
+      "step": 2358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405525320812994e-05,
+      "loss": 0.5847,
+      "step": 2359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050405922720536e-05,
+      "loss": 0.5698,
+      "step": 2360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045557498579e-05,
+      "loss": 0.6643,
+      "step": 2361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040707937068423e-05,
+      "loss": 0.7322,
+      "step": 2362
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4035857239552208e-05,
+      "loss": 0.585,
+      "step": 2363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031005407394078e-05,
+      "loss": 0.5977,
+      "step": 2364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026152441958085e-05,
+      "loss": 0.6821,
+      "step": 2365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021298344608589e-05,
+      "loss": 0.6113,
+      "step": 2366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016443116710274e-05,
+      "loss": 0.4873,
+      "step": 2367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011586759628139e-05,
+      "loss": 0.6519,
+      "step": 2368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006729274727505e-05,
+      "loss": 0.6384,
+      "step": 2369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001870663374e-05,
+      "loss": 0.5145,
+      "step": 2370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997010926933574e-05,
+      "loss": 0.6066,
+      "step": 2371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992150066772505e-05,
+      "loss": 0.4703,
+      "step": 2372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987288084257369e-05,
+      "loss": 0.5466,
+      "step": 2373
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982424980755064e-05,
+      "loss": 0.5996,
+      "step": 2374
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977560757632813e-05,
+      "loss": 0.7202,
+      "step": 2375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397269541625813e-05,
+      "loss": 0.5283,
+      "step": 2376
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967828957998873e-05,
+      "loss": 0.5498,
+      "step": 2377
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962961384223191e-05,
+      "loss": 0.5298,
+      "step": 2378
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958092696299554e-05,
+      "loss": 0.6139,
+      "step": 2379
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953222895596752e-05,
+      "loss": 0.6165,
+      "step": 2380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394835198348388e-05,
+      "loss": 0.4844,
+      "step": 2381
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943479961330344e-05,
+      "loss": 0.6792,
+      "step": 2382
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938606830505864e-05,
+      "loss": 0.6357,
+      "step": 2383
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933732592380485e-05,
+      "loss": 0.5918,
+      "step": 2384
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392885724832454e-05,
+      "loss": 0.5266,
+      "step": 2385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923980799708687e-05,
+      "loss": 0.6052,
+      "step": 2386
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919103247903898e-05,
+      "loss": 0.5748,
+      "step": 2387
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391422459428144e-05,
+      "loss": 0.6411,
+      "step": 2388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909344840212903e-05,
+      "loss": 0.6458,
+      "step": 2389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904463987070185e-05,
+      "loss": 0.5188,
+      "step": 2390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899582036225486e-05,
+      "loss": 0.522,
+      "step": 2391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894698989051321e-05,
+      "loss": 0.5134,
+      "step": 2392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889814846920509e-05,
+      "loss": 0.6147,
+      "step": 2393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884929611206182e-05,
+      "loss": 0.4675,
+      "step": 2394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880043283281774e-05,
+      "loss": 0.5192,
+      "step": 2395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 0.6313,
+      "step": 2396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870267356297999e-05,
+      "loss": 0.5481,
+      "step": 2397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386537775998704e-05,
+      "loss": 0.5128,
+      "step": 2398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860487076962815e-05,
+      "loss": 0.5979,
+      "step": 2399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855595308600284e-05,
+      "loss": 0.5862,
+      "step": 2400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850702456274732e-05,
+      "loss": 0.5015,
+      "step": 2401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845808521361727e-05,
+      "loss": 0.6306,
+      "step": 2402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840913505237157e-05,
+      "loss": 0.512,
+      "step": 2403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836017409277205e-05,
+      "loss": 0.5569,
+      "step": 2404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831120234858362e-05,
+      "loss": 0.5835,
+      "step": 2405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826221983357422e-05,
+      "loss": 0.5857,
+      "step": 2406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821322656151475e-05,
+      "loss": 0.5245,
+      "step": 2407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816422254617925e-05,
+      "loss": 0.5798,
+      "step": 2408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811520780134471e-05,
+      "loss": 0.5715,
+      "step": 2409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806618234079114e-05,
+      "loss": 0.5811,
+      "step": 2410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380171461783016e-05,
+      "loss": 0.4939,
+      "step": 2411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796809932766206e-05,
+      "loss": 0.6541,
+      "step": 2412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379190418026616e-05,
+      "loss": 0.5835,
+      "step": 2413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786997361709229e-05,
+      "loss": 0.6327,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782089478474915e-05,
+      "loss": 0.5798,
+      "step": 2415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777180531943021e-05,
+      "loss": 0.5793,
+      "step": 2416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377227052349365e-05,
+      "loss": 0.5125,
+      "step": 2417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767359454507201e-05,
+      "loss": 0.4437,
+      "step": 2418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762447326364377e-05,
+      "loss": 0.6978,
+      "step": 2419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757534140446174e-05,
+      "loss": 0.5835,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752619898133884e-05,
+      "loss": 0.5454,
+      "step": 2421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747704600809102e-05,
+      "loss": 0.5183,
+      "step": 2422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742788249853713e-05,
+      "loss": 0.6804,
+      "step": 2423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37378708466499e-05,
+      "loss": 0.6567,
+      "step": 2424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732952392580143e-05,
+      "loss": 0.5784,
+      "step": 2425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728032889027224e-05,
+      "loss": 0.687,
+      "step": 2426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723112337374204e-05,
+      "loss": 0.646,
+      "step": 2427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718190739004455e-05,
+      "loss": 0.5867,
+      "step": 2428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713268095301636e-05,
+      "loss": 0.6685,
+      "step": 2429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708344407649694e-05,
+      "loss": 0.562,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703419677432884e-05,
+      "loss": 0.5195,
+      "step": 2431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369849390603574e-05,
+      "loss": 0.6096,
+      "step": 2432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693567094843096e-05,
+      "loss": 0.6296,
+      "step": 2433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688639245240078e-05,
+      "loss": 0.5122,
+      "step": 2434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683710358612108e-05,
+      "loss": 0.7732,
+      "step": 2435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678780436344881e-05,
+      "loss": 0.5923,
+      "step": 2436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673849479824409e-05,
+      "loss": 0.5569,
+      "step": 2437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668917490436974e-05,
+      "loss": 0.5945,
+      "step": 2438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663984469569165e-05,
+      "loss": 0.75,
+      "step": 2439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659050418607844e-05,
+      "loss": 0.5461,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365411533894018e-05,
+      "loss": 0.6748,
+      "step": 2441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649179231953614e-05,
+      "loss": 0.6694,
+      "step": 2442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644242099035889e-05,
+      "loss": 0.7122,
+      "step": 2443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639303941575031e-05,
+      "loss": 0.6819,
+      "step": 2444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634364760959357e-05,
+      "loss": 0.5872,
+      "step": 2445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629424558577467e-05,
+      "loss": 0.644,
+      "step": 2446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362448333581825e-05,
+      "loss": 0.6028,
+      "step": 2447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619541094070887e-05,
+      "loss": 0.696,
+      "step": 2448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614597834724834e-05,
+      "loss": 0.4773,
+      "step": 2449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609653559169846e-05,
+      "loss": 0.637,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604708268795957e-05,
+      "loss": 0.5542,
+      "step": 2451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599761964993486e-05,
+      "loss": 0.5764,
+      "step": 2452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594814649153034e-05,
+      "loss": 0.5305,
+      "step": 2453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589866322665497e-05,
+      "loss": 0.5815,
+      "step": 2454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584916986922048e-05,
+      "loss": 0.6929,
+      "step": 2455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579966643314141e-05,
+      "loss": 0.6577,
+      "step": 2456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357501529323352e-05,
+      "loss": 0.5403,
+      "step": 2457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570062938072208e-05,
+      "loss": 0.689,
+      "step": 2458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565109579222511e-05,
+      "loss": 0.5625,
+      "step": 2459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560155218077017e-05,
+      "loss": 0.5922,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555199856028597e-05,
+      "loss": 0.5249,
+      "step": 2461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550243494470403e-05,
+      "loss": 0.5647,
+      "step": 2462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545286134795867e-05,
+      "loss": 0.6653,
+      "step": 2463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540327778398703e-05,
+      "loss": 0.7061,
+      "step": 2464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535368426672903e-05,
+      "loss": 0.478,
+      "step": 2465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530408081012742e-05,
+      "loss": 0.46,
+      "step": 2466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525446742812773e-05,
+      "loss": 0.5657,
+      "step": 2467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520484413467827e-05,
+      "loss": 0.6846,
+      "step": 2468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515521094373016e-05,
+      "loss": 0.6472,
+      "step": 2469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510556786923727e-05,
+      "loss": 0.5183,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505591492515625e-05,
+      "loss": 0.5364,
+      "step": 2471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350062521254466e-05,
+      "loss": 0.5439,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495657948407049e-05,
+      "loss": 0.6624,
+      "step": 2473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349068970149929e-05,
+      "loss": 0.5837,
+      "step": 2474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 0.5288,
+      "step": 2475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480750264960698e-05,
+      "loss": 0.4868,
+      "step": 2476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475779078124242e-05,
+      "loss": 0.5706,
+      "step": 2477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347080691410639e-05,
+      "loss": 0.657,
+      "step": 2478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465833774305016e-05,
+      "loss": 0.6504,
+      "step": 2479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460859660118271e-05,
+      "loss": 0.5518,
+      "step": 2480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455884572944577e-05,
+      "loss": 0.6272,
+      "step": 2481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450908514182632e-05,
+      "loss": 0.6548,
+      "step": 2482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445931485231407e-05,
+      "loss": 0.7039,
+      "step": 2483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440953487490145e-05,
+      "loss": 0.6934,
+      "step": 2484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435974522358361e-05,
+      "loss": 0.5836,
+      "step": 2485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430994591235844e-05,
+      "loss": 0.6023,
+      "step": 2486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342601369552265e-05,
+      "loss": 0.5594,
+      "step": 2487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421031836619111e-05,
+      "loss": 0.5706,
+      "step": 2488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416049015925828e-05,
+      "loss": 0.7334,
+      "step": 2489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411065234843673e-05,
+      "loss": 0.6206,
+      "step": 2490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406080494773785e-05,
+      "loss": 0.6868,
+      "step": 2491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340109479711758e-05,
+      "loss": 0.5635,
+      "step": 2492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396108143276733e-05,
+      "loss": 0.5586,
+      "step": 2493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391120534653198e-05,
+      "loss": 0.5144,
+      "step": 2494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386131972649187e-05,
+      "loss": 0.5913,
+      "step": 2495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381142458667192e-05,
+      "loss": 0.595,
+      "step": 2496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376151994109958e-05,
+      "loss": 0.5425,
+      "step": 2497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371160580380513e-05,
+      "loss": 0.6461,
+      "step": 2498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366168218882143e-05,
+      "loss": 0.6414,
+      "step": 2499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361174911018398e-05,
+      "loss": 0.4905,
+      "step": 2500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33561806581931e-05,
+      "loss": 0.5713,
+      "step": 2501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351185461810334e-05,
+      "loss": 0.6382,
+      "step": 2502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346189323274452e-05,
+      "loss": 0.5696,
+      "step": 2503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341192243990068e-05,
+      "loss": 0.7671,
+      "step": 2504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336194225362067e-05,
+      "loss": 0.5056,
+      "step": 2505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331195268795587e-05,
+      "loss": 0.5232,
+      "step": 2506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326195375696039e-05,
+      "loss": 0.5686,
+      "step": 2507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321194547469089e-05,
+      "loss": 0.5447,
+      "step": 2508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331619278552068e-05,
+      "loss": 0.5828,
+      "step": 2509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311190091257002e-05,
+      "loss": 0.687,
+      "step": 2510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306186466084516e-05,
+      "loss": 0.6244,
+      "step": 2511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301181911409944e-05,
+      "loss": 0.5699,
+      "step": 2512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296176428640262e-05,
+      "loss": 0.5391,
+      "step": 2513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329117001918272e-05,
+      "loss": 0.5862,
+      "step": 2514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286162684444816e-05,
+      "loss": 0.6958,
+      "step": 2515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281154425834315e-05,
+      "loss": 0.4596,
+      "step": 2516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276145244759241e-05,
+      "loss": 0.5632,
+      "step": 2517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271135142627878e-05,
+      "loss": 0.6494,
+      "step": 2518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266124120848763e-05,
+      "loss": 0.6499,
+      "step": 2519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261112180830696e-05,
+      "loss": 0.5885,
+      "step": 2520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256099323982739e-05,
+      "loss": 0.4971,
+      "step": 2521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251085551714207e-05,
+      "loss": 0.532,
+      "step": 2522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324607086543467e-05,
+      "loss": 0.5073,
+      "step": 2523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324105526655396e-05,
+      "loss": 0.6838,
+      "step": 2524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236038756482161e-05,
+      "loss": 0.6128,
+      "step": 2525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231021336629623e-05,
+      "loss": 0.5723,
+      "step": 2526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3226003008406941e-05,
+      "loss": 0.6123,
+      "step": 2527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220983773224967e-05,
+      "loss": 0.5129,
+      "step": 2528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321596363249481e-05,
+      "loss": 0.6125,
+      "step": 2529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210942587627837e-05,
+      "loss": 0.5157,
+      "step": 2530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205920640035666e-05,
+      "loss": 0.4565,
+      "step": 2531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200897791130163e-05,
+      "loss": 0.624,
+      "step": 2532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3195874042323457e-05,
+      "loss": 0.5581,
+      "step": 2533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190849395027926e-05,
+      "loss": 0.6235,
+      "step": 2534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185823850656201e-05,
+      "loss": 0.5569,
+      "step": 2535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180797410621162e-05,
+      "loss": 0.5908,
+      "step": 2536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175770076335946e-05,
+      "loss": 0.5327,
+      "step": 2537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317074184921394e-05,
+      "loss": 0.5305,
+      "step": 2538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165712730668778e-05,
+      "loss": 0.582,
+      "step": 2539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316068272211435e-05,
+      "loss": 0.4922,
+      "step": 2540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155651824964792e-05,
+      "loss": 0.3992,
+      "step": 2541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150620040634494e-05,
+      "loss": 0.6045,
+      "step": 2542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145587370538091e-05,
+      "loss": 0.5408,
+      "step": 2543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140553816090472e-05,
+      "loss": 0.5459,
+      "step": 2544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135519378706767e-05,
+      "loss": 0.5281,
+      "step": 2545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130484059802364e-05,
+      "loss": 0.553,
+      "step": 2546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312544786079289e-05,
+      "loss": 0.7559,
+      "step": 2547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120410783094224e-05,
+      "loss": 0.5747,
+      "step": 2548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115372828122493e-05,
+      "loss": 0.5878,
+      "step": 2549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110333997294067e-05,
+      "loss": 0.6958,
+      "step": 2550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105294292025566e-05,
+      "loss": 0.6177,
+      "step": 2551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100253713733855e-05,
+      "loss": 0.6506,
+      "step": 2552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095212263836039e-05,
+      "loss": 0.675,
+      "step": 2553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.5447,
+      "step": 2554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085126754891764e-05,
+      "loss": 0.552,
+      "step": 2555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080082698680747e-05,
+      "loss": 0.5901,
+      "step": 2556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307503777653451e-05,
+      "loss": 0.6238,
+      "step": 2557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069991989871387e-05,
+      "loss": 0.5797,
+      "step": 2558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306494534010995e-05,
+      "loss": 0.5798,
+      "step": 2559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059897828669014e-05,
+      "loss": 0.554,
+      "step": 2560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305484945696764e-05,
+      "loss": 0.5886,
+      "step": 2561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049800226425129e-05,
+      "loss": 0.5273,
+      "step": 2562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044750138461019e-05,
+      "loss": 0.5554,
+      "step": 2563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30396991944951e-05,
+      "loss": 0.5251,
+      "step": 2564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034647395947387e-05,
+      "loss": 0.6575,
+      "step": 2565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302959474423815e-05,
+      "loss": 0.5621,
+      "step": 2566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302454124078789e-05,
+      "loss": 0.5602,
+      "step": 2567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019486887017352e-05,
+      "loss": 0.5171,
+      "step": 2568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014431684347515e-05,
+      "loss": 0.6062,
+      "step": 2569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009375634199603e-05,
+      "loss": 0.45,
+      "step": 2570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004318737995076e-05,
+      "loss": 0.6528,
+      "step": 2571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999260997155632e-05,
+      "loss": 0.521,
+      "step": 2572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994202413103197e-05,
+      "loss": 0.4561,
+      "step": 2573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298914298725995e-05,
+      "loss": 0.5386,
+      "step": 2574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984082721048301e-05,
+      "loss": 0.6406,
+      "step": 2575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297902161589089e-05,
+      "loss": 0.6245,
+      "step": 2576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973959673210598e-05,
+      "loss": 0.5811,
+      "step": 2577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2968896894430546e-05,
+      "loss": 0.6384,
+      "step": 2578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963833280974073e-05,
+      "loss": 0.5192,
+      "step": 2579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958768834264777e-05,
+      "loss": 0.5417,
+      "step": 2580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953703555726471e-05,
+      "loss": 0.6743,
+      "step": 2581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948637446783213e-05,
+      "loss": 0.6167,
+      "step": 2582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943570508859287e-05,
+      "loss": 0.5576,
+      "step": 2583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938502743379212e-05,
+      "loss": 0.5281,
+      "step": 2584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293343415176774e-05,
+      "loss": 0.5806,
+      "step": 2585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928364735449863e-05,
+      "loss": 0.7017,
+      "step": 2586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923294495850793e-05,
+      "loss": 0.5701,
+      "step": 2587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291822343439598e-05,
+      "loss": 0.6985,
+      "step": 2588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913151552511102e-05,
+      "loss": 0.7017,
+      "step": 2589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908078851622071e-05,
+      "loss": 0.5806,
+      "step": 2590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903005333155023e-05,
+      "loss": 0.583,
+      "step": 2591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897930998536334e-05,
+      "loss": 0.5312,
+      "step": 2592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892855849192602e-05,
+      "loss": 0.6816,
+      "step": 2593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887779886550652e-05,
+      "loss": 0.616,
+      "step": 2594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882703112037547e-05,
+      "loss": 0.5779,
+      "step": 2595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877625527080568e-05,
+      "loss": 0.6293,
+      "step": 2596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872547133107229e-05,
+      "loss": 0.6172,
+      "step": 2597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867467931545275e-05,
+      "loss": 0.5593,
+      "step": 2598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862387923822672e-05,
+      "loss": 0.5969,
+      "step": 2599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857307111367611e-05,
+      "loss": 0.5518,
+      "step": 2600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285222549560852e-05,
+      "loss": 0.5986,
+      "step": 2601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284714307797404e-05,
+      "loss": 0.6191,
+      "step": 2602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842059859893043e-05,
+      "loss": 0.7031,
+      "step": 2603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836975842794632e-05,
+      "loss": 0.5808,
+      "step": 2604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831891028108121e-05,
+      "loss": 0.6812,
+      "step": 2605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826805417263062e-05,
+      "loss": 0.5376,
+      "step": 2606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821719011689222e-05,
+      "loss": 0.5728,
+      "step": 2607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816631812816594e-05,
+      "loss": 0.6304,
+      "step": 2608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811543822075396e-05,
+      "loss": 0.4915,
+      "step": 2609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806455040896069e-05,
+      "loss": 0.488,
+      "step": 2610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280136547070927e-05,
+      "loss": 0.7263,
+      "step": 2611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796275112945884e-05,
+      "loss": 0.5559,
+      "step": 2612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791183969037016e-05,
+      "loss": 0.6768,
+      "step": 2613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278609204041399e-05,
+      "loss": 0.532,
+      "step": 2614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780999328508348e-05,
+      "loss": 0.5898,
+      "step": 2615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277590583475187e-05,
+      "loss": 0.5732,
+      "step": 2616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770811560576526e-05,
+      "loss": 0.4512,
+      "step": 2617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765716507414532e-05,
+      "loss": 0.5574,
+      "step": 2618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760620676698307e-05,
+      "loss": 0.6523,
+      "step": 2619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755524069860497e-05,
+      "loss": 0.6367,
+      "step": 2620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750426688333964e-05,
+      "loss": 0.563,
+      "step": 2621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745328533551784e-05,
+      "loss": 0.5771,
+      "step": 2622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740229606947257e-05,
+      "loss": 0.6108,
+      "step": 2623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735129909953894e-05,
+      "loss": 0.6238,
+      "step": 2624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730029444005428e-05,
+      "loss": 0.4674,
+      "step": 2625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724928210535801e-05,
+      "loss": 0.6467,
+      "step": 2626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719826210979174e-05,
+      "loss": 0.5491,
+      "step": 2627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714723446769932e-05,
+      "loss": 0.5605,
+      "step": 2628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709619919342662e-05,
+      "loss": 0.5249,
+      "step": 2629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270451563013217e-05,
+      "loss": 0.5426,
+      "step": 2630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699410580573482e-05,
+      "loss": 0.7065,
+      "step": 2631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694304772101827e-05,
+      "loss": 0.678,
+      "step": 2632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 0.4291,
+      "step": 2633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684090884161636e-05,
+      "loss": 0.6741,
+      "step": 2634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267898280756463e-05,
+      "loss": 0.5359,
+      "step": 2635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673873977797731e-05,
+      "loss": 0.5791,
+      "step": 2636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668764396297235e-05,
+      "loss": 0.5251,
+      "step": 2637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663654064499651e-05,
+      "loss": 0.4875,
+      "step": 2638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658542983841695e-05,
+      "loss": 0.6221,
+      "step": 2639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653431155760307e-05,
+      "loss": 0.5994,
+      "step": 2640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264831858169262e-05,
+      "loss": 0.5465,
+      "step": 2641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643205263075985e-05,
+      "loss": 0.5552,
+      "step": 2642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638091201347967e-05,
+      "loss": 0.5522,
+      "step": 2643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632976397946325e-05,
+      "loss": 0.5759,
+      "step": 2644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627860854309044e-05,
+      "loss": 0.5999,
+      "step": 2645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622744571874305e-05,
+      "loss": 0.5403,
+      "step": 2646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617627552080506e-05,
+      "loss": 0.5815,
+      "step": 2647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612509796366243e-05,
+      "loss": 0.4719,
+      "step": 2648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607391306170324e-05,
+      "loss": 0.6011,
+      "step": 2649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602272082931764e-05,
+      "loss": 0.5549,
+      "step": 2650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597152128089778e-05,
+      "loss": 0.5898,
+      "step": 2651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25920314430838e-05,
+      "loss": 0.6831,
+      "step": 2652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586910029353451e-05,
+      "loss": 0.676,
+      "step": 2653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581787888338573e-05,
+      "loss": 0.519,
+      "step": 2654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576665021479204e-05,
+      "loss": 0.582,
+      "step": 2655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571541430215586e-05,
+      "loss": 0.687,
+      "step": 2656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566417115988169e-05,
+      "loss": 0.4666,
+      "step": 2657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561292080237606e-05,
+      "loss": 0.5977,
+      "step": 2658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556166324404747e-05,
+      "loss": 0.5083,
+      "step": 2659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2551039849930653e-05,
+      "loss": 0.5488,
+      "step": 2660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545912658256575e-05,
+      "loss": 0.6257,
+      "step": 2661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254078475082398e-05,
+      "loss": 0.5696,
+      "step": 2662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535656129074525e-05,
+      "loss": 0.5649,
+      "step": 2663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2530526794450074e-05,
+      "loss": 0.5986,
+      "step": 2664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252539674839269e-05,
+      "loss": 0.6135,
+      "step": 2665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520265992344633e-05,
+      "loss": 0.5686,
+      "step": 2666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515134527748365e-05,
+      "loss": 0.5999,
+      "step": 2667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251000235604655e-05,
+      "loss": 0.6753,
+      "step": 2668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2504869478682045e-05,
+      "loss": 0.5601,
+      "step": 2669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499735897097911e-05,
+      "loss": 0.515,
+      "step": 2670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494601612737403e-05,
+      "loss": 0.6177,
+      "step": 2671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489466627043976e-05,
+      "loss": 0.5381,
+      "step": 2672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484330941461278e-05,
+      "loss": 0.6479,
+      "step": 2673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479194557433163e-05,
+      "loss": 0.8022,
+      "step": 2674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474057476403666e-05,
+      "loss": 0.5461,
+      "step": 2675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468919699817041e-05,
+      "loss": 0.5204,
+      "step": 2676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246378122911771e-05,
+      "loss": 0.6802,
+      "step": 2677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458642065750312e-05,
+      "loss": 0.5959,
+      "step": 2678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245350221115967e-05,
+      "loss": 0.4529,
+      "step": 2679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448361666790803e-05,
+      "loss": 0.8042,
+      "step": 2680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443220434088925e-05,
+      "loss": 0.5139,
+      "step": 2681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438078514499447e-05,
+      "loss": 0.752,
+      "step": 2682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432935909467968e-05,
+      "loss": 0.6637,
+      "step": 2683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242779262044028e-05,
+      "loss": 0.6387,
+      "step": 2684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422648648862367e-05,
+      "loss": 0.5618,
+      "step": 2685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417503996180408e-05,
+      "loss": 0.5275,
+      "step": 2686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412358663840776e-05,
+      "loss": 0.6987,
+      "step": 2687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407212653290029e-05,
+      "loss": 0.6985,
+      "step": 2688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402065965974912e-05,
+      "loss": 0.9102,
+      "step": 2689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396918603342372e-05,
+      "loss": 0.4946,
+      "step": 2690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391770566839536e-05,
+      "loss": 0.5962,
+      "step": 2691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386621857913726e-05,
+      "loss": 0.6067,
+      "step": 2692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381472478012453e-05,
+      "loss": 0.5596,
+      "step": 2693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376322428583413e-05,
+      "loss": 0.5271,
+      "step": 2694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371171711074493e-05,
+      "loss": 0.4675,
+      "step": 2695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366020326933767e-05,
+      "loss": 0.6274,
+      "step": 2696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360868277609495e-05,
+      "loss": 0.6216,
+      "step": 2697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355715564550126e-05,
+      "loss": 0.5066,
+      "step": 2698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350562189204297e-05,
+      "loss": 0.5823,
+      "step": 2699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345408153020828e-05,
+      "loss": 0.5718,
+      "step": 2700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340253457448728e-05,
+      "loss": 0.7197,
+      "step": 2701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335098103937181e-05,
+      "loss": 0.501,
+      "step": 2702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329942093935575e-05,
+      "loss": 0.5986,
+      "step": 2703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324785428893464e-05,
+      "loss": 0.6499,
+      "step": 2704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319628110260603e-05,
+      "loss": 0.4991,
+      "step": 2705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231447013948691e-05,
+      "loss": 0.6206,
+      "step": 2706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230931151802251e-05,
+      "loss": 0.6138,
+      "step": 2707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304152247317692e-05,
+      "loss": 0.5625,
+      "step": 2708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298992328822937e-05,
+      "loss": 0.6284,
+      "step": 2709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293831763988901e-05,
+      "loss": 0.5908,
+      "step": 2710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288670554266435e-05,
+      "loss": 0.6052,
+      "step": 2711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 0.5801,
+      "step": 2712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278346205960476e-05,
+      "loss": 0.7034,
+      "step": 2713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273183070279574e-05,
+      "loss": 0.7036,
+      "step": 2714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268019295515411e-05,
+      "loss": 0.6855,
+      "step": 2715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226285488311974e-05,
+      "loss": 0.5518,
+      "step": 2716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257689834544486e-05,
+      "loss": 0.8457,
+      "step": 2717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252524151241745e-05,
+      "loss": 0.5287,
+      "step": 2718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247357834663804e-05,
+      "loss": 0.605,
+      "step": 2719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242190886263121e-05,
+      "loss": 0.6128,
+      "step": 2720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237023307492333e-05,
+      "loss": 0.7056,
+      "step": 2721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231855099804251e-05,
+      "loss": 0.6096,
+      "step": 2722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226686264651873e-05,
+      "loss": 0.6953,
+      "step": 2723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221516803488361e-05,
+      "loss": 0.5271,
+      "step": 2724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2216346717767064e-05,
+      "loss": 0.5608,
+      "step": 2725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2211176008941497e-05,
+      "loss": 0.6052,
+      "step": 2726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206004678465351e-05,
+      "loss": 0.5955,
+      "step": 2727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200832727792502e-05,
+      "loss": 0.5745,
+      "step": 2728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195660158376992e-05,
+      "loss": 0.5532,
+      "step": 2729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190486971673037e-05,
+      "loss": 0.6197,
+      "step": 2730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185313169135026e-05,
+      "loss": 0.6592,
+      "step": 2731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218013875221753e-05,
+      "loss": 0.6758,
+      "step": 2732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174963722375278e-05,
+      "loss": 0.5701,
+      "step": 2733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169788081063181e-05,
+      "loss": 0.6401,
+      "step": 2734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164611829736323e-05,
+      "loss": 0.5581,
+      "step": 2735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159434969849957e-05,
+      "loss": 0.4958,
+      "step": 2736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154257502859506e-05,
+      "loss": 0.6113,
+      "step": 2737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149079430220563e-05,
+      "loss": 0.5728,
+      "step": 2738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143900753388888e-05,
+      "loss": 0.6262,
+      "step": 2739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138721473820422e-05,
+      "loss": 0.5488,
+      "step": 2740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213354159297127e-05,
+      "loss": 0.5286,
+      "step": 2741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128361112297703e-05,
+      "loss": 0.5454,
+      "step": 2742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123180033256158e-05,
+      "loss": 0.5327,
+      "step": 2743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117998357303252e-05,
+      "loss": 0.5029,
+      "step": 2744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112816085895757e-05,
+      "loss": 0.6523,
+      "step": 2745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107633220490622e-05,
+      "loss": 0.5103,
+      "step": 2746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102449762544957e-05,
+      "loss": 0.4995,
+      "step": 2747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097265713516043e-05,
+      "loss": 0.4347,
+      "step": 2748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092081074861324e-05,
+      "loss": 0.7178,
+      "step": 2749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086895848038407e-05,
+      "loss": 0.5078,
+      "step": 2750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208171003450507e-05,
+      "loss": 0.6689,
+      "step": 2751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2076523635719255e-05,
+      "loss": 0.5791,
+      "step": 2752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207133665313907e-05,
+      "loss": 0.6511,
+      "step": 2753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066149088222778e-05,
+      "loss": 0.7554,
+      "step": 2754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060960942428819e-05,
+      "loss": 0.517,
+      "step": 2755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055772217215784e-05,
+      "loss": 0.5134,
+      "step": 2756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050582914042437e-05,
+      "loss": 0.5427,
+      "step": 2757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045393034367693e-05,
+      "loss": 0.5027,
+      "step": 2758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040202579650649e-05,
+      "loss": 0.6328,
+      "step": 2759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035011551350538e-05,
+      "loss": 0.656,
+      "step": 2760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029819950926776e-05,
+      "loss": 0.7134,
+      "step": 2761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024627779838922e-05,
+      "loss": 0.55,
+      "step": 2762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201943503954671e-05,
+      "loss": 0.6372,
+      "step": 2763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014241731510026e-05,
+      "loss": 0.5695,
+      "step": 2764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009047857188923e-05,
+      "loss": 0.5535,
+      "step": 2765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20038534180436e-05,
+      "loss": 0.6289,
+      "step": 2766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998658415534429e-05,
+      "loss": 0.5857,
+      "step": 2767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993462851121933e-05,
+      "loss": 0.5349,
+      "step": 2768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988266726266793e-05,
+      "loss": 0.6082,
+      "step": 2769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983070042429849e-05,
+      "loss": 0.4985,
+      "step": 2770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977872801072099e-05,
+      "loss": 0.5752,
+      "step": 2771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972675003654696e-05,
+      "loss": 0.5381,
+      "step": 2772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1967476651638956e-05,
+      "loss": 0.4844,
+      "step": 2773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962277746486334e-05,
+      "loss": 0.6184,
+      "step": 2774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957078289658455e-05,
+      "loss": 0.584,
+      "step": 2775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19518782826171e-05,
+      "loss": 0.5222,
+      "step": 2776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946677726824199e-05,
+      "loss": 0.6743,
+      "step": 2777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941476623741835e-05,
+      "loss": 0.5256,
+      "step": 2778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936274974832247e-05,
+      "loss": 0.6665,
+      "step": 2779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931072781557832e-05,
+      "loss": 0.5898,
+      "step": 2780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925870045381132e-05,
+      "loss": 0.5632,
+      "step": 2781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920666767764847e-05,
+      "loss": 0.5273,
+      "step": 2782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915462950171828e-05,
+      "loss": 0.6025,
+      "step": 2783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910258594065079e-05,
+      "loss": 0.5989,
+      "step": 2784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905053700907753e-05,
+      "loss": 0.7109,
+      "step": 2785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899848272163152e-05,
+      "loss": 0.6482,
+      "step": 2786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894642309294732e-05,
+      "loss": 0.5986,
+      "step": 2787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889435813766103e-05,
+      "loss": 0.5146,
+      "step": 2788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188422878704102e-05,
+      "loss": 0.679,
+      "step": 2789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1879021230583383e-05,
+      "loss": 0.5217,
+      "step": 2790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.5994,
+      "step": 2791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868604534326819e-05,
+      "loss": 0.7124,
+      "step": 2792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863395397456443e-05,
+      "loss": 0.5957,
+      "step": 2793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858185736710621e-05,
+      "loss": 0.6836,
+      "step": 2794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852975553554e-05,
+      "loss": 0.6943,
+      "step": 2795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847764849451369e-05,
+      "loss": 0.6243,
+      "step": 2796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842553625867668e-05,
+      "loss": 0.6389,
+      "step": 2797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837341884267978e-05,
+      "loss": 0.5266,
+      "step": 2798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832129626117533e-05,
+      "loss": 0.6172,
+      "step": 2799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826916852881709e-05,
+      "loss": 0.5732,
+      "step": 2800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821703566026024e-05,
+      "loss": 0.5811,
+      "step": 2801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816489767016145e-05,
+      "loss": 0.5503,
+      "step": 2802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811275457317879e-05,
+      "loss": 0.6216,
+      "step": 2803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806060638397181e-05,
+      "loss": 0.5676,
+      "step": 2804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800845311720141e-05,
+      "loss": 0.5844,
+      "step": 2805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795629478753004e-05,
+      "loss": 0.5879,
+      "step": 2806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790413140962144e-05,
+      "loss": 0.6528,
+      "step": 2807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178519629981409e-05,
+      "loss": 0.564,
+      "step": 2808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779978956775507e-05,
+      "loss": 0.5139,
+      "step": 2809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774761113313189e-05,
+      "loss": 0.438,
+      "step": 2810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769542770894092e-05,
+      "loss": 0.5659,
+      "step": 2811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764323930985298e-05,
+      "loss": 0.4758,
+      "step": 2812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759104595054034e-05,
+      "loss": 0.5464,
+      "step": 2813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753884764567665e-05,
+      "loss": 0.4667,
+      "step": 2814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748664440993695e-05,
+      "loss": 0.452,
+      "step": 2815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174344362579977e-05,
+      "loss": 0.6709,
+      "step": 2816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738222320453668e-05,
+      "loss": 0.5389,
+      "step": 2817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733000526423309e-05,
+      "loss": 0.6299,
+      "step": 2818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727778245176749e-05,
+      "loss": 0.542,
+      "step": 2819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722555478182183e-05,
+      "loss": 0.5261,
+      "step": 2820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717332226907942e-05,
+      "loss": 0.572,
+      "step": 2821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712108492822488e-05,
+      "loss": 0.5675,
+      "step": 2822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706884277394426e-05,
+      "loss": 0.6787,
+      "step": 2823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701659582092493e-05,
+      "loss": 0.4775,
+      "step": 2824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169643440838556e-05,
+      "loss": 0.5205,
+      "step": 2825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691208757742633e-05,
+      "loss": 0.6384,
+      "step": 2826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685982631632857e-05,
+      "loss": 0.5461,
+      "step": 2827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680756031525502e-05,
+      "loss": 0.7261,
+      "step": 2828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675528958889974e-05,
+      "loss": 0.6294,
+      "step": 2829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167030141519582e-05,
+      "loss": 0.6543,
+      "step": 2830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665073401912708e-05,
+      "loss": 0.6777,
+      "step": 2831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659844920510445e-05,
+      "loss": 0.5776,
+      "step": 2832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654615972458968e-05,
+      "loss": 0.468,
+      "step": 2833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649386559228342e-05,
+      "loss": 0.5635,
+      "step": 2834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644156682288767e-05,
+      "loss": 0.6338,
+      "step": 2835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638926343110574e-05,
+      "loss": 0.6157,
+      "step": 2836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163369554316422e-05,
+      "loss": 0.5674,
+      "step": 2837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628464283920293e-05,
+      "loss": 0.6423,
+      "step": 2838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623232566849512e-05,
+      "loss": 0.6165,
+      "step": 2839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618000393422719e-05,
+      "loss": 0.5151,
+      "step": 2840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1612767765110893e-05,
+      "loss": 0.6338,
+      "step": 2841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1607534683385135e-05,
+      "loss": 0.5753,
+      "step": 2842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602301149716677e-05,
+      "loss": 0.5637,
+      "step": 2843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597067165576875e-05,
+      "loss": 0.5291,
+      "step": 2844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159183273243721e-05,
+      "loss": 0.541,
+      "step": 2845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158659785176929e-05,
+      "loss": 0.657,
+      "step": 2846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158136252504486e-05,
+      "loss": 0.4858,
+      "step": 2847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576126753735772e-05,
+      "loss": 0.6055,
+      "step": 2848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570890539314015e-05,
+      "loss": 0.5486,
+      "step": 2849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565653883251702e-05,
+      "loss": 0.5723,
+      "step": 2850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560416787021065e-05,
+      "loss": 0.6711,
+      "step": 2851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555179252094462e-05,
+      "loss": 0.5876,
+      "step": 2852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549941279944374e-05,
+      "loss": 0.6443,
+      "step": 2853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544702872043411e-05,
+      "loss": 0.6291,
+      "step": 2854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539464029864298e-05,
+      "loss": 0.4666,
+      "step": 2855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534224754879884e-05,
+      "loss": 0.5281,
+      "step": 2856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152898504856314e-05,
+      "loss": 0.5037,
+      "step": 2857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523744912387153e-05,
+      "loss": 0.5933,
+      "step": 2858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518504347825146e-05,
+      "loss": 0.4927,
+      "step": 2859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513263356350446e-05,
+      "loss": 0.521,
+      "step": 2860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508021939436508e-05,
+      "loss": 0.5955,
+      "step": 2861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502780098556908e-05,
+      "loss": 0.554,
+      "step": 2862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497537835185336e-05,
+      "loss": 0.6538,
+      "step": 2863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14922951507956e-05,
+      "loss": 0.6174,
+      "step": 2864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487052046861635e-05,
+      "loss": 0.5479,
+      "step": 2865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481808524857487e-05,
+      "loss": 0.4948,
+      "step": 2866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476564586257321e-05,
+      "loss": 0.5378,
+      "step": 2867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471320232535419e-05,
+      "loss": 0.7319,
+      "step": 2868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466075465166182e-05,
+      "loss": 0.553,
+      "step": 2869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 0.5974,
+      "step": 2870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455584695383865e-05,
+      "loss": 0.6465,
+      "step": 2871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1450338695920169e-05,
+      "loss": 0.5842,
+      "step": 2872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445092288707888e-05,
+      "loss": 0.6062,
+      "step": 2873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439845475222002e-05,
+      "loss": 0.427,
+      "step": 2874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14345982569376e-05,
+      "loss": 0.6355,
+      "step": 2875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429350635329884e-05,
+      "loss": 0.5276,
+      "step": 2876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424102611874175e-05,
+      "loss": 0.5886,
+      "step": 2877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418854188045904e-05,
+      "loss": 0.6833,
+      "step": 2878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413605365320613e-05,
+      "loss": 0.5491,
+      "step": 2879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408356145173955e-05,
+      "loss": 0.5527,
+      "step": 2880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14031065290817e-05,
+      "loss": 0.5564,
+      "step": 2881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397856518519726e-05,
+      "loss": 0.5723,
+      "step": 2882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392606114964019e-05,
+      "loss": 0.5454,
+      "step": 2883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387355319890685e-05,
+      "loss": 0.6064,
+      "step": 2884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382104134775929e-05,
+      "loss": 0.6289,
+      "step": 2885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137685256109607e-05,
+      "loss": 0.6243,
+      "step": 2886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371600600327538e-05,
+      "loss": 0.709,
+      "step": 2887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366348253946872e-05,
+      "loss": 0.6787,
+      "step": 2888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361095523430718e-05,
+      "loss": 0.4878,
+      "step": 2889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1355842410255831e-05,
+      "loss": 0.6609,
+      "step": 2890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350588915899069e-05,
+      "loss": 0.7334,
+      "step": 2891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13453350418374e-05,
+      "loss": 0.6755,
+      "step": 2892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340080789547905e-05,
+      "loss": 0.552,
+      "step": 2893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133482616050776e-05,
+      "loss": 0.5637,
+      "step": 2894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329571156194258e-05,
+      "loss": 0.7002,
+      "step": 2895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324315778084788e-05,
+      "loss": 0.6514,
+      "step": 2896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319060027656847e-05,
+      "loss": 0.6646,
+      "step": 2897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313803906388044e-05,
+      "loss": 0.5925,
+      "step": 2898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308547415756084e-05,
+      "loss": 0.7383,
+      "step": 2899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303290557238773e-05,
+      "loss": 0.4714,
+      "step": 2900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298033332314032e-05,
+      "loss": 0.4407,
+      "step": 2901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1292775742459875e-05,
+      "loss": 0.6023,
+      "step": 2902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287517789154425e-05,
+      "loss": 0.6138,
+      "step": 2903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282259473875903e-05,
+      "loss": 0.7019,
+      "step": 2904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277000798102631e-05,
+      "loss": 0.7295,
+      "step": 2905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271741763313037e-05,
+      "loss": 0.6121,
+      "step": 2906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266482370985647e-05,
+      "loss": 0.6213,
+      "step": 2907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261222622599086e-05,
+      "loss": 0.7734,
+      "step": 2908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255962519632082e-05,
+      "loss": 0.5422,
+      "step": 2909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250702063563464e-05,
+      "loss": 0.5269,
+      "step": 2910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245441255872154e-05,
+      "loss": 0.5083,
+      "step": 2911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240180098037177e-05,
+      "loss": 0.5264,
+      "step": 2912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123491859153766e-05,
+      "loss": 0.585,
+      "step": 2913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229656737852823e-05,
+      "loss": 0.6667,
+      "step": 2914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224394538461983e-05,
+      "loss": 0.615,
+      "step": 2915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219131994844556e-05,
+      "loss": 0.7368,
+      "step": 2916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213869108480058e-05,
+      "loss": 0.5757,
+      "step": 2917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208605880848095e-05,
+      "loss": 0.6985,
+      "step": 2918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203342313428374e-05,
+      "loss": 0.5742,
+      "step": 2919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198078407700695e-05,
+      "loss": 0.5479,
+      "step": 2920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192814165144952e-05,
+      "loss": 0.5096,
+      "step": 2921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187549587241137e-05,
+      "loss": 0.5792,
+      "step": 2922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182284675469335e-05,
+      "loss": 0.6367,
+      "step": 2923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177019431309724e-05,
+      "loss": 0.5757,
+      "step": 2924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171753856242577e-05,
+      "loss": 0.6724,
+      "step": 2925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116648795174826e-05,
+      "loss": 0.6377,
+      "step": 2926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161221719307226e-05,
+      "loss": 0.6792,
+      "step": 2927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155955160400032e-05,
+      "loss": 0.5591,
+      "step": 2928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150688276507313e-05,
+      "loss": 0.7534,
+      "step": 2929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145421069109807e-05,
+      "loss": 0.5443,
+      "step": 2930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140153539688333e-05,
+      "loss": 0.5479,
+      "step": 2931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113488568972381e-05,
+      "loss": 0.5232,
+      "step": 2932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129617520697244e-05,
+      "loss": 0.5964,
+      "step": 2933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124349034089724e-05,
+      "loss": 0.5757,
+      "step": 2934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119080231382435e-05,
+      "loss": 0.6221,
+      "step": 2935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111381111405665e-05,
+      "loss": 0.5591,
+      "step": 2936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108541683593731e-05,
+      "loss": 0.5833,
+      "step": 2937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103271941475129e-05,
+      "loss": 0.6621,
+      "step": 2938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098001889182378e-05,
+      "loss": 0.6013,
+      "step": 2939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092731528197101e-05,
+      "loss": 0.6161,
+      "step": 2940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087460860001009e-05,
+      "loss": 0.4266,
+      "step": 2941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082189886075898e-05,
+      "loss": 0.7363,
+      "step": 2942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076918607903655e-05,
+      "loss": 0.5852,
+      "step": 2943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071647026966245e-05,
+      "loss": 0.595,
+      "step": 2944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066375144745725e-05,
+      "loss": 0.5403,
+      "step": 2945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061102962724228e-05,
+      "loss": 0.7002,
+      "step": 2946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105583048238398e-05,
+      "loss": 0.6443,
+      "step": 2947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050557705207283e-05,
+      "loss": 0.5627,
+      "step": 2948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.6748,
+      "step": 2949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040011266274205e-05,
+      "loss": 0.6396,
+      "step": 2950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034737607482845e-05,
+      "loss": 0.5962,
+      "step": 2951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10294636577851e-05,
+      "loss": 0.5273,
+      "step": 2952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024189418663685e-05,
+      "loss": 0.8179,
+      "step": 2953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018914891601397e-05,
+      "loss": 0.5027,
+      "step": 2954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013640078081127e-05,
+      "loss": 0.7588,
+      "step": 2955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100836497958583e-05,
+      "loss": 0.6479,
+      "step": 2956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003089597598552e-05,
+      "loss": 0.5232,
+      "step": 2957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997813933602415e-05,
+      "loss": 0.5127,
+      "step": 2958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992537989080618e-05,
+      "loss": 0.5837,
+      "step": 2959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987261765516442e-05,
+      "loss": 0.554,
+      "step": 2960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981985264393249e-05,
+      "loss": 0.759,
+      "step": 2961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976708487194474e-05,
+      "loss": 0.5542,
+      "step": 2962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0971431435403632e-05,
+      "loss": 0.6091,
+      "step": 2963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966154110504312e-05,
+      "loss": 0.6792,
+      "step": 2964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960876513980185e-05,
+      "loss": 0.6108,
+      "step": 2965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955598647314996e-05,
+      "loss": 0.5781,
+      "step": 2966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950320511992565e-05,
+      "loss": 0.4897,
+      "step": 2967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945042109496789e-05,
+      "loss": 0.5665,
+      "step": 2968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939763441311637e-05,
+      "loss": 0.4744,
+      "step": 2969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934484508921157e-05,
+      "loss": 0.645,
+      "step": 2970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929205313809465e-05,
+      "loss": 0.4685,
+      "step": 2971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923925857460764e-05,
+      "loss": 0.5496,
+      "step": 2972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918646141359318e-05,
+      "loss": 0.7483,
+      "step": 2973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913366166989462e-05,
+      "loss": 0.605,
+      "step": 2974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908085935835618e-05,
+      "loss": 0.5369,
+      "step": 2975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902805449382265e-05,
+      "loss": 0.6602,
+      "step": 2976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897524709113965e-05,
+      "loss": 0.5387,
+      "step": 2977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892243716515344e-05,
+      "loss": 0.6082,
+      "step": 2978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886962473071101e-05,
+      "loss": 0.5691,
+      "step": 2979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881680980266007e-05,
+      "loss": 0.7305,
+      "step": 2980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876399239584903e-05,
+      "loss": 0.46,
+      "step": 2981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08711172525127e-05,
+      "loss": 0.5054,
+      "step": 2982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086583502053437e-05,
+      "loss": 0.5869,
+      "step": 2983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086055254513497e-05,
+      "loss": 0.5874,
+      "step": 2984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855269827799613e-05,
+      "loss": 0.6777,
+      "step": 2985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849986870013481e-05,
+      "loss": 0.5952,
+      "step": 2986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844703673261829e-05,
+      "loss": 0.6375,
+      "step": 2987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839420239029978e-05,
+      "loss": 0.5378,
+      "step": 2988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083413656880331e-05,
+      "loss": 0.6753,
+      "step": 2989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828852664067279e-05,
+      "loss": 0.6462,
+      "step": 2990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823568526307405e-05,
+      "loss": 0.6211,
+      "step": 2991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818284157009267e-05,
+      "loss": 0.6753,
+      "step": 2992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812999557658519e-05,
+      "loss": 0.5188,
+      "step": 2993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080771472974087e-05,
+      "loss": 0.5708,
+      "step": 2994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802429674742102e-05,
+      "loss": 0.6143,
+      "step": 2995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797144394148053e-05,
+      "loss": 0.5361,
+      "step": 2996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791858889444633e-05,
+      "loss": 0.5549,
+      "step": 2997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786573162117809e-05,
+      "loss": 0.5715,
+      "step": 2998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781287213653606e-05,
+      "loss": 0.7126,
+      "step": 2999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776001045538122e-05,
+      "loss": 0.6733,
+      "step": 3000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770714659257513e-05,
+      "loss": 0.605,
+      "step": 3001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765428056297991e-05,
+      "loss": 0.4404,
+      "step": 3002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760141238145833e-05,
+      "loss": 0.6836,
+      "step": 3003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075485420628738e-05,
+      "loss": 0.7161,
+      "step": 3004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0749566962209023e-05,
+      "loss": 0.549,
+      "step": 3005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744279507397229e-05,
+      "loss": 0.5974,
+      "step": 3006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0738991843338502e-05,
+      "loss": 0.5579,
+      "step": 3007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733703971519426e-05,
+      "loss": 0.6477,
+      "step": 3008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728415893426636e-05,
+      "loss": 0.5605,
+      "step": 3009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723127610546818e-05,
+      "loss": 0.5511,
+      "step": 3010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717839124366723e-05,
+      "loss": 0.6152,
+      "step": 3011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712550436373158e-05,
+      "loss": 0.573,
+      "step": 3012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707261548052985e-05,
+      "loss": 0.6721,
+      "step": 3013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701972460893124e-05,
+      "loss": 0.6536,
+      "step": 3014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069668317638055e-05,
+      "loss": 0.4912,
+      "step": 3015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691393696002295e-05,
+      "loss": 0.5114,
+      "step": 3016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686104021245443e-05,
+      "loss": 0.4035,
+      "step": 3017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0680814153597134e-05,
+      "loss": 0.5343,
+      "step": 3018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675524094544564e-05,
+      "loss": 0.5237,
+      "step": 3019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670233845574982e-05,
+      "loss": 0.6384,
+      "step": 3020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664943408175692e-05,
+      "loss": 0.5503,
+      "step": 3021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659652783834045e-05,
+      "loss": 0.5444,
+      "step": 3022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654361974037448e-05,
+      "loss": 0.5051,
+      "step": 3023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0649070980273363e-05,
+      "loss": 0.5901,
+      "step": 3024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643779804029304e-05,
+      "loss": 0.688,
+      "step": 3025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638488446792828e-05,
+      "loss": 0.5925,
+      "step": 3026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063319691005155e-05,
+      "loss": 0.6667,
+      "step": 3027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.5959,
+      "step": 3028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622613304005298e-05,
+      "loss": 0.5114,
+      "step": 3029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617321237675802e-05,
+      "loss": 0.5945,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612028997792455e-05,
+      "loss": 0.5789,
+      "step": 3031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606736585843124e-05,
+      "loss": 0.5515,
+      "step": 3032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601444003315723e-05,
+      "loss": 0.5396,
+      "step": 3033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05961512516982e-05,
+      "loss": 0.6123,
+      "step": 3034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590858332478566e-05,
+      "loss": 0.6008,
+      "step": 3035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585565247144876e-05,
+      "loss": 0.4639,
+      "step": 3036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058027199718522e-05,
+      "loss": 0.5978,
+      "step": 3037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574978584087758e-05,
+      "loss": 0.4641,
+      "step": 3038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569685009340668e-05,
+      "loss": 0.5557,
+      "step": 3039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564391274432196e-05,
+      "loss": 0.5972,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559097380850618e-05,
+      "loss": 0.6873,
+      "step": 3041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553803330084264e-05,
+      "loss": 0.6357,
+      "step": 3042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05485091236215e-05,
+      "loss": 0.4978,
+      "step": 3043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543214762950747e-05,
+      "loss": 0.5321,
+      "step": 3044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537920249560461e-05,
+      "loss": 0.4788,
+      "step": 3045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053262558493914e-05,
+      "loss": 0.6018,
+      "step": 3046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052733077057533e-05,
+      "loss": 0.4956,
+      "step": 3047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522035807957616e-05,
+      "loss": 0.6907,
+      "step": 3048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516740698574624e-05,
+      "loss": 0.7063,
+      "step": 3049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511445443915025e-05,
+      "loss": 0.5232,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506150045467525e-05,
+      "loss": 0.7046,
+      "step": 3051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500854504720874e-05,
+      "loss": 0.5554,
+      "step": 3052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495558823163865e-05,
+      "loss": 0.5784,
+      "step": 3053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490263002285321e-05,
+      "loss": 0.5669,
+      "step": 3054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484967043574114e-05,
+      "loss": 0.644,
+      "step": 3055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479670948519157e-05,
+      "loss": 0.5303,
+      "step": 3056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474374718609386e-05,
+      "loss": 0.6814,
+      "step": 3057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469078355333788e-05,
+      "loss": 0.6655,
+      "step": 3058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463781860181385e-05,
+      "loss": 0.6375,
+      "step": 3059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458485234641236e-05,
+      "loss": 0.5117,
+      "step": 3060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045318848020243e-05,
+      "loss": 0.4875,
+      "step": 3061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447891598354104e-05,
+      "loss": 0.7075,
+      "step": 3062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442594590585422e-05,
+      "loss": 0.5808,
+      "step": 3063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437297458385587e-05,
+      "loss": 0.5699,
+      "step": 3064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432000203243836e-05,
+      "loss": 0.6443,
+      "step": 3065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426702826649435e-05,
+      "loss": 0.5679,
+      "step": 3066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421405330091701e-05,
+      "loss": 0.5522,
+      "step": 3067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416107715059965e-05,
+      "loss": 0.6233,
+      "step": 3068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04108099830436e-05,
+      "loss": 0.6665,
+      "step": 3069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405512135532019e-05,
+      "loss": 0.7231,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400214174014651e-05,
+      "loss": 0.7073,
+      "step": 3071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394916099980972e-05,
+      "loss": 0.5889,
+      "step": 3072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389617914920481e-05,
+      "loss": 0.5991,
+      "step": 3073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384319620322712e-05,
+      "loss": 0.4807,
+      "step": 3074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0379021217677233e-05,
+      "loss": 0.7605,
+      "step": 3075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037372270847363e-05,
+      "loss": 0.717,
+      "step": 3076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368424094201535e-05,
+      "loss": 0.6243,
+      "step": 3077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363125376350593e-05,
+      "loss": 0.5264,
+      "step": 3078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357826556410495e-05,
+      "loss": 0.5914,
+      "step": 3079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035252763587095e-05,
+      "loss": 0.5071,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347228616221698e-05,
+      "loss": 0.7932,
+      "step": 3081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341929498952506e-05,
+      "loss": 0.5959,
+      "step": 3082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033663028555317e-05,
+      "loss": 0.5376,
+      "step": 3083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033133097751351e-05,
+      "loss": 0.5205,
+      "step": 3084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326031576323374e-05,
+      "loss": 0.5945,
+      "step": 3085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320732083472642e-05,
+      "loss": 0.6133,
+      "step": 3086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315432500451215e-05,
+      "loss": 0.5347,
+      "step": 3087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310132828749013e-05,
+      "loss": 0.5947,
+      "step": 3088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030483306985599e-05,
+      "loss": 0.6897,
+      "step": 3089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029953322526212e-05,
+      "loss": 0.5229,
+      "step": 3090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294233296457407e-05,
+      "loss": 0.6085,
+      "step": 3091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028893328493187e-05,
+      "loss": 0.6252,
+      "step": 3092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028363319217556e-05,
+      "loss": 0.5823,
+      "step": 3093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278333019678542e-05,
+      "loss": 0.4866,
+      "step": 3094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273032768930908e-05,
+      "loss": 0.583,
+      "step": 3095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267732441422775e-05,
+      "loss": 0.5555,
+      "step": 3096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262432038644275e-05,
+      "loss": 0.4819,
+      "step": 3097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257131562085567e-05,
+      "loss": 0.4915,
+      "step": 3098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025183101323683e-05,
+      "loss": 0.6323,
+      "step": 3099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0246530393588256e-05,
+      "loss": 0.7178,
+      "step": 3100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241229704630068e-05,
+      "loss": 0.5815,
+      "step": 3101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235928947852499e-05,
+      "loss": 0.5411,
+      "step": 3102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230628124745805e-05,
+      "loss": 0.5278,
+      "step": 3103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225327236800263e-05,
+      "loss": 0.604,
+      "step": 3104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022002628550617e-05,
+      "loss": 0.5996,
+      "step": 3105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021472527235383e-05,
+      "loss": 0.7275,
+      "step": 3106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.6018,
+      "step": 3107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204123066435742e-05,
+      "loss": 0.5938,
+      "step": 3108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198821876650702e-05,
+      "loss": 0.6074,
+      "step": 3109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193520630968832e-05,
+      "loss": 0.6072,
+      "step": 3110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188219330880524e-05,
+      "loss": 0.5884,
+      "step": 3111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182917977876185e-05,
+      "loss": 0.488,
+      "step": 3112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177616573446241e-05,
+      "loss": 0.6042,
+      "step": 3113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172315119081127e-05,
+      "loss": 0.4763,
+      "step": 3114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167013616271298e-05,
+      "loss": 0.644,
+      "step": 3115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016171206650722e-05,
+      "loss": 0.6516,
+      "step": 3116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156410471279373e-05,
+      "loss": 0.6064,
+      "step": 3117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151108832078245e-05,
+      "loss": 0.5562,
+      "step": 3118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145807150394341e-05,
+      "loss": 0.5886,
+      "step": 3119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140505427718181e-05,
+      "loss": 0.7642,
+      "step": 3120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135203665540288e-05,
+      "loss": 0.6375,
+      "step": 3121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129901865351204e-05,
+      "loss": 0.5432,
+      "step": 3122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124600028641478e-05,
+      "loss": 0.7651,
+      "step": 3123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119298156901669e-05,
+      "loss": 0.5452,
+      "step": 3124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113996251622344e-05,
+      "loss": 0.6448,
+      "step": 3125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108694314294085e-05,
+      "loss": 0.4528,
+      "step": 3126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103392346407478e-05,
+      "loss": 0.5133,
+      "step": 3127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098090349453124e-05,
+      "loss": 0.5549,
+      "step": 3128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092788324921625e-05,
+      "loss": 0.5281,
+      "step": 3129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087486274303591e-05,
+      "loss": 0.6885,
+      "step": 3130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082184199089645e-05,
+      "loss": 0.6194,
+      "step": 3131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076882100770413e-05,
+      "loss": 0.6787,
+      "step": 3132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071579980836526e-05,
+      "loss": 0.5781,
+      "step": 3133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0066277840778626e-05,
+      "loss": 0.6187,
+      "step": 3134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006097568208736e-05,
+      "loss": 0.6294,
+      "step": 3135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055673506253377e-05,
+      "loss": 0.6438,
+      "step": 3136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050371314767327e-05,
+      "loss": 0.531,
+      "step": 3137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004506910911987e-05,
+      "loss": 0.5298,
+      "step": 3138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039766890801676e-05,
+      "loss": 0.7334,
+      "step": 3139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003446466130341e-05,
+      "loss": 0.5142,
+      "step": 3140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029162422115737e-05,
+      "loss": 0.6562,
+      "step": 3141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023860174729336e-05,
+      "loss": 0.5527,
+      "step": 3142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018557920634877e-05,
+      "loss": 0.6746,
+      "step": 3143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001325566132304e-05,
+      "loss": 0.6172,
+      "step": 3144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007953398284505e-05,
+      "loss": 0.5378,
+      "step": 3145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002651133009949e-05,
+      "loss": 0.5059,
+      "step": 3146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997348866990053e-06,
+      "loss": 0.6028,
+      "step": 3147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992046601715498e-06,
+      "loss": 0.575,
+      "step": 3148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98674433867696e-06,
+      "loss": 0.6968,
+      "step": 3149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981442079365124e-06,
+      "loss": 0.572,
+      "step": 3150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976139825270666e-06,
+      "loss": 0.5195,
+      "step": 3151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970837577884267e-06,
+      "loss": 0.6289,
+      "step": 3152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965535338696597e-06,
+      "loss": 0.5688,
+      "step": 3153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960233109198327e-06,
+      "loss": 0.6294,
+      "step": 3154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954930890880132e-06,
+      "loss": 0.6069,
+      "step": 3155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949628685232677e-06,
+      "loss": 0.6819,
+      "step": 3156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94432649374663e-06,
+      "loss": 0.5513,
+      "step": 3157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93902431791264e-06,
+      "loss": 0.7087,
+      "step": 3158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933722159221375e-06,
+      "loss": 0.74,
+      "step": 3159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928420019163474e-06,
+      "loss": 0.6477,
+      "step": 3160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923117899229589e-06,
+      "loss": 0.605,
+      "step": 3161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917815800910358e-06,
+      "loss": 0.6594,
+      "step": 3162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91251372569641e-06,
+      "loss": 0.5552,
+      "step": 3163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907211675078377e-06,
+      "loss": 0.5676,
+      "step": 3164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901909650546878e-06,
+      "loss": 0.5618,
+      "step": 3165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896607653592527e-06,
+      "loss": 0.5569,
+      "step": 3166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891305685705919e-06,
+      "loss": 0.5679,
+      "step": 3167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886003748377661e-06,
+      "loss": 0.6875,
+      "step": 3168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880701843098335e-06,
+      "loss": 0.4614,
+      "step": 3169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875399971358527e-06,
+      "loss": 0.6831,
+      "step": 3170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870098134648798e-06,
+      "loss": 0.5103,
+      "step": 3171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864796334459716e-06,
+      "loss": 0.6589,
+      "step": 3172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859494572281822e-06,
+      "loss": 0.6599,
+      "step": 3173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85419284960566e-06,
+      "loss": 0.5962,
+      "step": 3174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848891167921756e-06,
+      "loss": 0.5718,
+      "step": 3175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843589528720632e-06,
+      "loss": 0.6921,
+      "step": 3176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838287933492785e-06,
+      "loss": 0.5315,
+      "step": 3177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832986383728706e-06,
+      "loss": 0.6885,
+      "step": 3178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827684880918877e-06,
+      "loss": 0.6375,
+      "step": 3179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822383426553762e-06,
+      "loss": 0.7632,
+      "step": 3180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81708202212382e-06,
+      "loss": 0.5833,
+      "step": 3181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811780669119479e-06,
+      "loss": 0.572,
+      "step": 3182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806479369031172e-06,
+      "loss": 0.6301,
+      "step": 3183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801178123349298e-06,
+      "loss": 0.571,
+      "step": 3184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79587693356426e-06,
+      "loss": 0.6528,
+      "step": 3185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 0.5947,
+      "step": 3186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785274727646174e-06,
+      "loss": 0.5095,
+      "step": 3187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779973714493832e-06,
+      "loss": 0.5592,
+      "step": 3188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774672763199738e-06,
+      "loss": 0.5669,
+      "step": 3189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769371875254197e-06,
+      "loss": 0.4963,
+      "step": 3190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764071052147506e-06,
+      "loss": 0.542,
+      "step": 3191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758770295369937e-06,
+      "loss": 0.5896,
+      "step": 3192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753469606411745e-06,
+      "loss": 0.6658,
+      "step": 3193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748168986763175e-06,
+      "loss": 0.5562,
+      "step": 3194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742868437914433e-06,
+      "loss": 0.5388,
+      "step": 3195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737567961355728e-06,
+      "loss": 0.6172,
+      "step": 3196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732267558577227e-06,
+      "loss": 0.5236,
+      "step": 3197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726967231069094e-06,
+      "loss": 0.5571,
+      "step": 3198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72166698032146e-06,
+      "loss": 0.6147,
+      "step": 3199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716366807824444e-06,
+      "loss": 0.5337,
+      "step": 3200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711066715068134e-06,
+      "loss": 0.6533,
+      "step": 3201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705766703542596e-06,
+      "loss": 0.6946,
+      "step": 3202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700466774737882e-06,
+      "loss": 0.6433,
+      "step": 3203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695166930144013e-06,
+      "loss": 0.5784,
+      "step": 3204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689867171250993e-06,
+      "loss": 0.5417,
+      "step": 3205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684567499548789e-06,
+      "loss": 0.5452,
+      "step": 3206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679267916527361e-06,
+      "loss": 0.5403,
+      "step": 3207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673968423676627e-06,
+      "loss": 0.4814,
+      "step": 3208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668669022486495e-06,
+      "loss": 0.6077,
+      "step": 3209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663369714446832e-06,
+      "loss": 0.6025,
+      "step": 3210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658070501047498e-06,
+      "loss": 0.6611,
+      "step": 3211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652771383778303e-06,
+      "loss": 0.6663,
+      "step": 3212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647472364129053e-06,
+      "loss": 0.4875,
+      "step": 3213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642173443589508e-06,
+      "loss": 0.5759,
+      "step": 3214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63687462364941e-06,
+      "loss": 0.623,
+      "step": 3215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631575905798471e-06,
+      "loss": 0.5652,
+      "step": 3216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626277291526372e-06,
+      "loss": 0.5073,
+      "step": 3217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620978782322772e-06,
+      "loss": 0.6031,
+      "step": 3218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61568037967729e-06,
+      "loss": 0.6854,
+      "step": 3219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610382085079522e-06,
+      "loss": 0.7219,
+      "step": 3220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60508390001903e-06,
+      "loss": 0.5991,
+      "step": 3221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599785825985352e-06,
+      "loss": 0.6604,
+      "step": 3222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594487864467983e-06,
+      "loss": 0.5631,
+      "step": 3223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5891900169564e-06,
+      "loss": 0.7812,
+      "step": 3224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58389228494004e-06,
+      "loss": 0.4191,
+      "step": 3225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578594669908304e-06,
+      "loss": 0.5142,
+      "step": 3226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573297173350567e-06,
+      "loss": 0.635,
+      "step": 3227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567999796756168e-06,
+      "loss": 0.5581,
+      "step": 3228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562702541614416e-06,
+      "loss": 0.5486,
+      "step": 3229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55740540941458e-06,
+      "loss": 0.5161,
+      "step": 3230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5521084016459e-06,
+      "loss": 0.5381,
+      "step": 3231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546811519797572e-06,
+      "loss": 0.6282,
+      "step": 3232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541514765358768e-06,
+      "loss": 0.5417,
+      "step": 3233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536218139818615e-06,
+      "loss": 0.6929,
+      "step": 3234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530921644666215e-06,
+      "loss": 0.6992,
+      "step": 3235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525625281390616e-06,
+      "loss": 0.5847,
+      "step": 3236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520329051480848e-06,
+      "loss": 0.5288,
+      "step": 3237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515032956425887e-06,
+      "loss": 0.6309,
+      "step": 3238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50973699771468e-06,
+      "loss": 0.7178,
+      "step": 3239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504441176836141e-06,
+      "loss": 0.6226,
+      "step": 3240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499145495279128e-06,
+      "loss": 0.5383,
+      "step": 3241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49384995453248e-06,
+      "loss": 0.5419,
+      "step": 3242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488554556084979e-06,
+      "loss": 0.511,
+      "step": 3243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48325930142538e-06,
+      "loss": 0.5806,
+      "step": 3244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477964192042386e-06,
+      "loss": 0.6399,
+      "step": 3245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472669229424673e-06,
+      "loss": 0.5613,
+      "step": 3246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467374415060861e-06,
+      "loss": 0.6672,
+      "step": 3247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462079750439542e-06,
+      "loss": 0.6191,
+      "step": 3248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456785237049253e-06,
+      "loss": 0.6616,
+      "step": 3249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451490876378502e-06,
+      "loss": 0.5361,
+      "step": 3250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44619666991574e-06,
+      "loss": 0.5974,
+      "step": 3251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440902619149385e-06,
+      "loss": 0.668,
+      "step": 3252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43560872556781e-06,
+      "loss": 0.5796,
+      "step": 3253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430314990659334e-06,
+      "loss": 0.5857,
+      "step": 3254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425021415912248e-06,
+      "loss": 0.5996,
+      "step": 3255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419728002814779e-06,
+      "loss": 0.6711,
+      "step": 3256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414434752855127e-06,
+      "loss": 0.5366,
+      "step": 3257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409141667521433e-06,
+      "loss": 0.5459,
+      "step": 3258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403848748301802e-06,
+      "loss": 0.6016,
+      "step": 3259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398555996684279e-06,
+      "loss": 0.6113,
+      "step": 3260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393263414156877e-06,
+      "loss": 0.5881,
+      "step": 3261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387971002207547e-06,
+      "loss": 0.5828,
+      "step": 3262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382678762324202e-06,
+      "loss": 0.6775,
+      "step": 3263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377386695994707e-06,
+      "loss": 0.7393,
+      "step": 3264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.6558,
+      "step": 3265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366803089948454e-06,
+      "loss": 0.6665,
+      "step": 3266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361511553207176e-06,
+      "loss": 0.5142,
+      "step": 3267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356220195970698e-06,
+      "loss": 0.5662,
+      "step": 3268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350929019726638e-06,
+      "loss": 0.5498,
+      "step": 3269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345638025962554e-06,
+      "loss": 0.6807,
+      "step": 3270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340347216165957e-06,
+      "loss": 0.6041,
+      "step": 3271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335056591824312e-06,
+      "loss": 0.5166,
+      "step": 3272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329766154425018e-06,
+      "loss": 0.5215,
+      "step": 3273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324475905455439e-06,
+      "loss": 0.543,
+      "step": 3274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319185846402871e-06,
+      "loss": 0.6301,
+      "step": 3275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313895978754562e-06,
+      "loss": 0.5947,
+      "step": 3276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308606303997711e-06,
+      "loss": 0.6013,
+      "step": 3277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303316823619452e-06,
+      "loss": 0.6431,
+      "step": 3278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298027539106883e-06,
+      "loss": 0.4637,
+      "step": 3279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292738451947018e-06,
+      "loss": 0.6477,
+      "step": 3280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287449563626846e-06,
+      "loss": 0.4918,
+      "step": 3281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282160875633279e-06,
+      "loss": 0.5862,
+      "step": 3282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276872389453185e-06,
+      "loss": 0.5901,
+      "step": 3283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271584106573364e-06,
+      "loss": 0.5354,
+      "step": 3284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266296028480575e-06,
+      "loss": 0.6364,
+      "step": 3285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2610081566615e-06,
+      "loss": 0.6282,
+      "step": 3286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255720492602776e-06,
+      "loss": 0.5891,
+      "step": 3287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25043303779098e-06,
+      "loss": 0.4905,
+      "step": 3288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245145793712625e-06,
+      "loss": 0.5869,
+      "step": 3289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239858761854172e-06,
+      "loss": 0.6257,
+      "step": 3290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234571943702012e-06,
+      "loss": 0.6179,
+      "step": 3291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22928534074249e-06,
+      "loss": 0.6243,
+      "step": 3292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22399895446188e-06,
+      "loss": 0.573,
+      "step": 3293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218712786346396e-06,
+      "loss": 0.5908,
+      "step": 3294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213426837882193e-06,
+      "loss": 0.5229,
+      "step": 3295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208141110555368e-06,
+      "loss": 0.666,
+      "step": 3296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202855605851945e-06,
+      "loss": 0.5913,
+      "step": 3297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197570325257902e-06,
+      "loss": 0.6567,
+      "step": 3298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192285270259134e-06,
+      "loss": 0.6108,
+      "step": 3299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187000442341485e-06,
+      "loss": 0.5494,
+      "step": 3300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181715842990738e-06,
+      "loss": 0.4656,
+      "step": 3301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1764314736926e-06,
+      "loss": 0.7192,
+      "step": 3302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171147335932726e-06,
+      "loss": 0.5073,
+      "step": 3303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165863431196693e-06,
+      "loss": 0.5786,
+      "step": 3304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160579760970025e-06,
+      "loss": 0.634,
+      "step": 3305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15529632673817e-06,
+      "loss": 0.6152,
+      "step": 3306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15001312998652e-06,
+      "loss": 0.5945,
+      "step": 3307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144730172200388e-06,
+      "loss": 0.5808,
+      "step": 3308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139447454865034e-06,
+      "loss": 0.7051,
+      "step": 3309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134164979465633e-06,
+      "loss": 0.5881,
+      "step": 3310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128882747487305e-06,
+      "loss": 0.5906,
+      "step": 3311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1236007604151e-06,
+      "loss": 0.4944,
+      "step": 3312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118319019733995e-06,
+      "loss": 0.5186,
+      "step": 3313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113037526928904e-06,
+      "loss": 0.5256,
+      "step": 3314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107756283484659e-06,
+      "loss": 0.5708,
+      "step": 3315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102475290886037e-06,
+      "loss": 0.656,
+      "step": 3316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097194550617733e-06,
+      "loss": 0.5884,
+      "step": 3317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091914064164384e-06,
+      "loss": 0.6016,
+      "step": 3318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086633833010538e-06,
+      "loss": 0.5994,
+      "step": 3319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081353858640687e-06,
+      "loss": 0.5984,
+      "step": 3320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076074142539236e-06,
+      "loss": 0.481,
+      "step": 3321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070794686190537e-06,
+      "loss": 0.5354,
+      "step": 3322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06551549107885e-06,
+      "loss": 0.5618,
+      "step": 3323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060236558688367e-06,
+      "loss": 0.6897,
+      "step": 3324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054957890503218e-06,
+      "loss": 0.3849,
+      "step": 3325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049679488007438e-06,
+      "loss": 0.5505,
+      "step": 3326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044401352685009e-06,
+      "loss": 0.647,
+      "step": 3327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039123486019817e-06,
+      "loss": 0.5234,
+      "step": 3328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033845889495691e-06,
+      "loss": 0.6738,
+      "step": 3329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02856856459637e-06,
+      "loss": 0.4493,
+      "step": 3330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02329151280553e-06,
+      "loss": 0.7173,
+      "step": 3331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01801473560675e-06,
+      "loss": 0.5737,
+      "step": 3332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01273823448356e-06,
+      "loss": 0.6748,
+      "step": 3333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007462010919387e-06,
+      "loss": 0.5168,
+      "step": 3334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002186066397588e-06,
+      "loss": 0.6913,
+      "step": 3335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996910402401453e-06,
+      "loss": 0.6152,
+      "step": 3336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991635020414173e-06,
+      "loss": 0.5508,
+      "step": 3337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986359921918878e-06,
+      "loss": 0.5911,
+      "step": 3338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981085108398604e-06,
+      "loss": 0.731,
+      "step": 3339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97581058133632e-06,
+      "loss": 0.5413,
+      "step": 3340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9705363422149e-06,
+      "loss": 0.5426,
+      "step": 3341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965262392517156e-06,
+      "loss": 0.6704,
+      "step": 3342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959988733725797e-06,
+      "loss": 0.6204,
+      "step": 3343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.6702,
+      "step": 3344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949442294792717e-06,
+      "loss": 0.5696,
+      "step": 3345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944169517616023e-06,
+      "loss": 0.6458,
+      "step": 3346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938897037275777e-06,
+      "loss": 0.6101,
+      "step": 3347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933624855254279e-06,
+      "loss": 0.656,
+      "step": 3348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92835297303376e-06,
+      "loss": 0.5564,
+      "step": 3349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923081392096348e-06,
+      "loss": 0.5114,
+      "step": 3350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917810113924105e-06,
+      "loss": 0.6082,
+      "step": 3351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912539139998994e-06,
+      "loss": 0.6213,
+      "step": 3352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907268471802904e-06,
+      "loss": 0.624,
+      "step": 3353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901998110817624e-06,
+      "loss": 0.7021,
+      "step": 3354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896728058524873e-06,
+      "loss": 0.6711,
+      "step": 3355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891458316406269e-06,
+      "loss": 0.6838,
+      "step": 3356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886188885943352e-06,
+      "loss": 0.6233,
+      "step": 3357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88091976861757e-06,
+      "loss": 0.6292,
+      "step": 3358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87565096591028e-06,
+      "loss": 0.5002,
+      "step": 3359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870382479302761e-06,
+      "loss": 0.6729,
+      "step": 3360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86511431027619e-06,
+      "loss": 0.6843,
+      "step": 3361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859846460311672e-06,
+      "loss": 0.5796,
+      "step": 3362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854578930890196e-06,
+      "loss": 0.5356,
+      "step": 3363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849311723492689e-06,
+      "loss": 0.5367,
+      "step": 3364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84404483959997e-06,
+      "loss": 0.5881,
+      "step": 3365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838778280692776e-06,
+      "loss": 0.5818,
+      "step": 3366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.833512048251742e-06,
+      "loss": 0.6523,
+      "step": 3367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828246143757426e-06,
+      "loss": 0.5758,
+      "step": 3368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822980568690278e-06,
+      "loss": 0.6082,
+      "step": 3369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817715324530667e-06,
+      "loss": 0.6345,
+      "step": 3370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812450412758868e-06,
+      "loss": 0.6378,
+      "step": 3371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807185834855053e-06,
+      "loss": 0.5447,
+      "step": 3372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801921592299312e-06,
+      "loss": 0.5659,
+      "step": 3373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79665768657163e-06,
+      "loss": 0.5881,
+      "step": 3374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791394119151908e-06,
+      "loss": 0.6482,
+      "step": 3375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786130891519945e-06,
+      "loss": 0.5439,
+      "step": 3376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780868005155446e-06,
+      "loss": 0.54,
+      "step": 3377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775605461538019e-06,
+      "loss": 0.6101,
+      "step": 3378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77034326214718e-06,
+      "loss": 0.6016,
+      "step": 3379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76508140846234e-06,
+      "loss": 0.6912,
+      "step": 3380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759819901962826e-06,
+      "loss": 0.5916,
+      "step": 3381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75455874412785e-06,
+      "loss": 0.5247,
+      "step": 3382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749297936436539e-06,
+      "loss": 0.6694,
+      "step": 3383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744037480367922e-06,
+      "loss": 0.5452,
+      "step": 3384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738777377400917e-06,
+      "loss": 0.5403,
+      "step": 3385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733517629014358e-06,
+      "loss": 0.5168,
+      "step": 3386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728258236686966e-06,
+      "loss": 0.6545,
+      "step": 3387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722999201897372e-06,
+      "loss": 0.5469,
+      "step": 3388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717740526124099e-06,
+      "loss": 0.6655,
+      "step": 3389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712482210845578e-06,
+      "loss": 0.5513,
+      "step": 3390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707224257540126e-06,
+      "loss": 0.7324,
+      "step": 3391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701966667685972e-06,
+      "loss": 0.5645,
+      "step": 3392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696709442761229e-06,
+      "loss": 0.5427,
+      "step": 3393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69145258424392e-06,
+      "loss": 0.5398,
+      "step": 3394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686196093611961e-06,
+      "loss": 0.4685,
+      "step": 3395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680939972343156e-06,
+      "loss": 0.5309,
+      "step": 3396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675684221915219e-06,
+      "loss": 0.6289,
+      "step": 3397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670428843805746e-06,
+      "loss": 0.6023,
+      "step": 3398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665173839492242e-06,
+      "loss": 0.5442,
+      "step": 3399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659919210452097e-06,
+      "loss": 0.6499,
+      "step": 3400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654664958162601e-06,
+      "loss": 0.5935,
+      "step": 3401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649411084100933e-06,
+      "loss": 0.4739,
+      "step": 3402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644157589744174e-06,
+      "loss": 0.5828,
+      "step": 3403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638904476569282e-06,
+      "loss": 0.5654,
+      "step": 3404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63365174605313e-06,
+      "loss": 0.6121,
+      "step": 3405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628399399672465e-06,
+      "loss": 0.6477,
+      "step": 3406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623147438903932e-06,
+      "loss": 0.5651,
+      "step": 3407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617895865224078e-06,
+      "loss": 0.5698,
+      "step": 3408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61264468010932e-06,
+      "loss": 0.5457,
+      "step": 3409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607393885035986e-06,
+      "loss": 0.5261,
+      "step": 3410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602143481480278e-06,
+      "loss": 0.6467,
+      "step": 3411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596893470918304e-06,
+      "loss": 0.5847,
+      "step": 3412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591643854826047e-06,
+      "loss": 0.603,
+      "step": 3413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586394634679392e-06,
+      "loss": 0.6013,
+      "step": 3414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581145811954097e-06,
+      "loss": 0.5994,
+      "step": 3415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575897388125827e-06,
+      "loss": 0.5576,
+      "step": 3416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570649364670116e-06,
+      "loss": 0.4829,
+      "step": 3417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565401743062403e-06,
+      "loss": 0.4943,
+      "step": 3418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560154524778003e-06,
+      "loss": 0.6382,
+      "step": 3419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554907711292113e-06,
+      "loss": 0.6482,
+      "step": 3420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549661304079838e-06,
+      "loss": 0.55,
+      "step": 3421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544415304616138e-06,
+      "loss": 0.574,
+      "step": 3422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 0.5149,
+      "step": 3423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533924534833822e-06,
+      "loss": 0.5984,
+      "step": 3424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528679767464584e-06,
+      "loss": 0.5667,
+      "step": 3425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52343541374268e-06,
+      "loss": 0.5037,
+      "step": 3426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518191475142515e-06,
+      "loss": 0.6001,
+      "step": 3427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512947953138367e-06,
+      "loss": 0.6494,
+      "step": 3428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507704849204402e-06,
+      "loss": 0.4115,
+      "step": 3429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502462164814668e-06,
+      "loss": 0.6934,
+      "step": 3430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497219901443094e-06,
+      "loss": 0.4968,
+      "step": 3431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491978060563495e-06,
+      "loss": 0.7722,
+      "step": 3432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486736643649557e-06,
+      "loss": 0.5574,
+      "step": 3433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481495652174859e-06,
+      "loss": 0.5321,
+      "step": 3434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476255087612849e-06,
+      "loss": 0.6051,
+      "step": 3435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.471014951436865e-06,
+      "loss": 0.7954,
+      "step": 3436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465775245120118e-06,
+      "loss": 0.6321,
+      "step": 3437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460535970135705e-06,
+      "loss": 0.6888,
+      "step": 3438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455297127956589e-06,
+      "loss": 0.6895,
+      "step": 3439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450058720055628e-06,
+      "loss": 0.594,
+      "step": 3440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44482074790554e-06,
+      "loss": 0.6343,
+      "step": 3441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43958321297894e-06,
+      "loss": 0.6914,
+      "step": 3442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434346116748303e-06,
+      "loss": 0.7432,
+      "step": 3443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429109460685987e-06,
+      "loss": 0.553,
+      "step": 3444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423873246264233e-06,
+      "loss": 0.511,
+      "step": 3445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418637474955145e-06,
+      "loss": 0.6157,
+      "step": 3446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413402148230712e-06,
+      "loss": 0.5823,
+      "step": 3447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408167267562792e-06,
+      "loss": 0.6555,
+      "step": 3448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40293283442313e-06,
+      "loss": 0.6482,
+      "step": 3449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397698850283325e-06,
+      "loss": 0.6438,
+      "step": 3450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.392465316614867e-06,
+      "loss": 0.4985,
+      "step": 3451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.387232234889108e-06,
+      "loss": 0.6296,
+      "step": 3452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381999606577283e-06,
+      "loss": 0.5369,
+      "step": 3453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.376767433150491e-06,
+      "loss": 0.5625,
+      "step": 3454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37153571607971e-06,
+      "loss": 0.6863,
+      "step": 3455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366304456835786e-06,
+      "loss": 0.5762,
+      "step": 3456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361073656889427e-06,
+      "loss": 0.5857,
+      "step": 3457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355843317711235e-06,
+      "loss": 0.6104,
+      "step": 3458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350613440771661e-06,
+      "loss": 0.6072,
+      "step": 3459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345384027541035e-06,
+      "loss": 0.5999,
+      "step": 3460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340155079489557e-06,
+      "loss": 0.4695,
+      "step": 3461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334926598087296e-06,
+      "loss": 0.5166,
+      "step": 3462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329698584804182e-06,
+      "loss": 0.4788,
+      "step": 3463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32447104111003e-06,
+      "loss": 0.5825,
+      "step": 3464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319243968474501e-06,
+      "loss": 0.6294,
+      "step": 3465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314017368367146e-06,
+      "loss": 0.5813,
+      "step": 3466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30879124225737e-06,
+      "loss": 0.4944,
+      "step": 3467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303565591614443e-06,
+      "loss": 0.4736,
+      "step": 3468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298340417907512e-06,
+      "loss": 0.5117,
+      "step": 3469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293115722605577e-06,
+      "loss": 0.5076,
+      "step": 3470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287891507177514e-06,
+      "loss": 0.6541,
+      "step": 3471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282667773092059e-06,
+      "loss": 0.5889,
+      "step": 3472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277444521817819e-06,
+      "loss": 0.6008,
+      "step": 3473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272221754823253e-06,
+      "loss": 0.6006,
+      "step": 3474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266999473576696e-06,
+      "loss": 0.5586,
+      "step": 3475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261777679546334e-06,
+      "loss": 0.6292,
+      "step": 3476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256556374200233e-06,
+      "loss": 0.626,
+      "step": 3477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251335559006304e-06,
+      "loss": 0.5947,
+      "step": 3478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246115235432338e-06,
+      "loss": 0.5505,
+      "step": 3479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240895404945971e-06,
+      "loss": 0.6108,
+      "step": 3480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235676069014707e-06,
+      "loss": 0.603,
+      "step": 3481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230457229105913e-06,
+      "loss": 0.5461,
+      "step": 3482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225238886686814e-06,
+      "loss": 0.6365,
+      "step": 3483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2200210432245e-06,
+      "loss": 0.5498,
+      "step": 3484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21480370018591e-06,
+      "loss": 0.5649,
+      "step": 3485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209586859037857e-06,
+      "loss": 0.5886,
+      "step": 3486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204370521246998e-06,
+      "loss": 0.5121,
+      "step": 3487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199154688279862e-06,
+      "loss": 0.7607,
+      "step": 3488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193939361602824e-06,
+      "loss": 0.6084,
+      "step": 3489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188724542682123e-06,
+      "loss": 0.5713,
+      "step": 3490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18351023298386e-06,
+      "loss": 0.5725,
+      "step": 3491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17829643397398e-06,
+      "loss": 0.655,
+      "step": 3492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173083147118296e-06,
+      "loss": 0.7163,
+      "step": 3493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16787037388247e-06,
+      "loss": 0.7373,
+      "step": 3494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162658115732027e-06,
+      "loss": 0.6643,
+      "step": 3495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157446374132335e-06,
+      "loss": 0.7925,
+      "step": 3496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152235150548635e-06,
+      "loss": 0.6736,
+      "step": 3497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147024446446002e-06,
+      "loss": 0.5803,
+      "step": 3498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14181426328938e-06,
+      "loss": 0.5947,
+      "step": 3499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136604602543557e-06,
+      "loss": 0.6068,
+      "step": 3500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131395465673184e-06,
+      "loss": 0.6067,
+      "step": 3501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.4929,
+      "step": 3502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12097876941662e-06,
+      "loss": 0.5994,
+      "step": 3503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115771212958987e-06,
+      "loss": 0.5574,
+      "step": 3504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1105641862339e-06,
+      "loss": 0.5801,
+      "step": 3505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10535769070527e-06,
+      "loss": 0.6516,
+      "step": 3506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10015172783685e-06,
+      "loss": 0.5642,
+      "step": 3507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094946299092252e-06,
+      "loss": 0.4871,
+      "step": 3508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089741405934923e-06,
+      "loss": 0.4727,
+      "step": 3509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084537049828173e-06,
+      "loss": 0.5645,
+      "step": 3510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079333232235154e-06,
+      "loss": 0.5818,
+      "step": 3511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074129954618873e-06,
+      "loss": 0.6931,
+      "step": 3512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068927218442171e-06,
+      "loss": 0.4746,
+      "step": 3513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063725025167755e-06,
+      "loss": 0.6382,
+      "step": 3514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05852337625817e-06,
+      "loss": 0.6953,
+      "step": 3515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053322273175804e-06,
+      "loss": 0.5911,
+      "step": 3516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048121717382904e-06,
+      "loss": 0.5979,
+      "step": 3517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042921710341546e-06,
+      "loss": 0.6401,
+      "step": 3518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03772225351367e-06,
+      "loss": 0.5198,
+      "step": 3519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032523348361047e-06,
+      "loss": 0.6724,
+      "step": 3520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027324996345306e-06,
+      "loss": 0.4941,
+      "step": 3521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0221271989279e-06,
+      "loss": 0.5618,
+      "step": 3522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016929957570154e-06,
+      "loss": 0.4946,
+      "step": 3523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011733273733208e-06,
+      "loss": 0.7158,
+      "step": 3524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00653714887807e-06,
+      "loss": 0.6206,
+      "step": 3525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.00134158446557e-06,
+      "loss": 0.6448,
+      "step": 3526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996146581956403e-06,
+      "loss": 0.5601,
+      "step": 3527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990952142811084e-06,
+      "loss": 0.6113,
+      "step": 3528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.985758268489975e-06,
+      "loss": 0.646,
+      "step": 3529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980564960453294e-06,
+      "loss": 0.5745,
+      "step": 3530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97537222016108e-06,
+      "loss": 0.6367,
+      "step": 3531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97018004907323e-06,
+      "loss": 0.6282,
+      "step": 3532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964988448649464e-06,
+      "loss": 0.5183,
+      "step": 3533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959797420349356e-06,
+      "loss": 0.6846,
+      "step": 3534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954606965632305e-06,
+      "loss": 0.6338,
+      "step": 3535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949417085957566e-06,
+      "loss": 0.5659,
+      "step": 3536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944227782784219e-06,
+      "loss": 0.6108,
+      "step": 3537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939039057571186e-06,
+      "loss": 0.5874,
+      "step": 3538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933850911777227e-06,
+      "loss": 0.6379,
+      "step": 3539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928663346860933e-06,
+      "loss": 0.6577,
+      "step": 3540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923476364280748e-06,
+      "loss": 0.6528,
+      "step": 3541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918289965494935e-06,
+      "loss": 0.4666,
+      "step": 3542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913104151961598e-06,
+      "loss": 0.5769,
+      "step": 3543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90791892513868e-06,
+      "loss": 0.5698,
+      "step": 3544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90273428648396e-06,
+      "loss": 0.5593,
+      "step": 3545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897550237455043e-06,
+      "loss": 0.6599,
+      "step": 3546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892366779509381e-06,
+      "loss": 0.5952,
+      "step": 3547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887183914104245e-06,
+      "loss": 0.6968,
+      "step": 3548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88200164269675e-06,
+      "loss": 0.5938,
+      "step": 3549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876819966743842e-06,
+      "loss": 0.7749,
+      "step": 3550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871638887702302e-06,
+      "loss": 0.8035,
+      "step": 3551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866458407028733e-06,
+      "loss": 0.6704,
+      "step": 3552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86127852617958e-06,
+      "loss": 0.5852,
+      "step": 3553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856099246611115e-06,
+      "loss": 0.5994,
+      "step": 3554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85092056977944e-06,
+      "loss": 0.6025,
+      "step": 3555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845742497140498e-06,
+      "loss": 0.6196,
+      "step": 3556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840565030150043e-06,
+      "loss": 0.5483,
+      "step": 3557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835388170263678e-06,
+      "loss": 0.5698,
+      "step": 3558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83021191893682e-06,
+      "loss": 0.5952,
+      "step": 3559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825036277624726e-06,
+      "loss": 0.6543,
+      "step": 3560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819861247782475e-06,
+      "loss": 0.5233,
+      "step": 3561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814686830864975e-06,
+      "loss": 0.5825,
+      "step": 3562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80951302832697e-06,
+      "loss": 0.6003,
+      "step": 3563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804339841623011e-06,
+      "loss": 0.5112,
+      "step": 3564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7991672722075e-06,
+      "loss": 0.7266,
+      "step": 3565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793995321534652e-06,
+      "loss": 0.6118,
+      "step": 3566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78882399105851e-06,
+      "loss": 0.5466,
+      "step": 3567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78365328223294e-06,
+      "loss": 0.6013,
+      "step": 3568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77848319651164e-06,
+      "loss": 0.5961,
+      "step": 3569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773313735348129e-06,
+      "loss": 0.5789,
+      "step": 3570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768144900195752e-06,
+      "loss": 0.5629,
+      "step": 3571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76297669250767e-06,
+      "loss": 0.7122,
+      "step": 3572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757809113736882e-06,
+      "loss": 0.6631,
+      "step": 3573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752642165336195e-06,
+      "loss": 0.6631,
+      "step": 3574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747475848758257e-06,
+      "loss": 0.6162,
+      "step": 3575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74231016545552e-06,
+      "loss": 0.5544,
+      "step": 3576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737145116880261e-06,
+      "loss": 0.5028,
+      "step": 3577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731980704484592e-06,
+      "loss": 0.635,
+      "step": 3578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72681692972043e-06,
+      "loss": 0.656,
+      "step": 3579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72165379403953e-06,
+      "loss": 0.624,
+      "step": 3580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 0.6055,
+      "step": 3581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711329445733568e-06,
+      "loss": 0.5576,
+      "step": 3582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706168236011099e-06,
+      "loss": 0.559,
+      "step": 3583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701007671177066e-06,
+      "loss": 0.6167,
+      "step": 3584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695847752682308e-06,
+      "loss": 0.6238,
+      "step": 3585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690688481977493e-06,
+      "loss": 0.5991,
+      "step": 3586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685529860513093e-06,
+      "loss": 0.5696,
+      "step": 3587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6803718897394e-06,
+      "loss": 0.5803,
+      "step": 3588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675214571106538e-06,
+      "loss": 0.5632,
+      "step": 3589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670057906064429e-06,
+      "loss": 0.4814,
+      "step": 3590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664901896062822e-06,
+      "loss": 0.6021,
+      "step": 3591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659746542551275e-06,
+      "loss": 0.572,
+      "step": 3592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654591846979176e-06,
+      "loss": 0.6736,
+      "step": 3593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649437810795703e-06,
+      "loss": 0.8179,
+      "step": 3594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644284435449877e-06,
+      "loss": 0.6436,
+      "step": 3595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639131722390507e-06,
+      "loss": 0.7607,
+      "step": 3596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633979673066237e-06,
+      "loss": 0.5481,
+      "step": 3597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628828288925507e-06,
+      "loss": 0.7688,
+      "step": 3598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62367757141659e-06,
+      "loss": 0.4535,
+      "step": 3599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618527521987552e-06,
+      "loss": 0.6401,
+      "step": 3600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6133781420862774e-06,
+      "loss": 0.5385,
+      "step": 3601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608229433160468e-06,
+      "loss": 0.585,
+      "step": 3602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603081396657632e-06,
+      "loss": 0.5625,
+      "step": 3603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597934034025093e-06,
+      "loss": 0.6648,
+      "step": 3604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592787346709975e-06,
+      "loss": 0.5115,
+      "step": 3605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587641336159227e-06,
+      "loss": 0.5435,
+      "step": 3606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5824960038195915e-06,
+      "loss": 0.7114,
+      "step": 3607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577351351137635e-06,
+      "loss": 0.5223,
+      "step": 3608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572207379559722e-06,
+      "loss": 0.6221,
+      "step": 3609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567064090532035e-06,
+      "loss": 0.6504,
+      "step": 3610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561921485500553e-06,
+      "loss": 0.594,
+      "step": 3611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556779565911077e-06,
+      "loss": 0.5267,
+      "step": 3612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551638333209202e-06,
+      "loss": 0.7222,
+      "step": 3613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546497788840333e-06,
+      "loss": 0.5691,
+      "step": 3614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541357934249692e-06,
+      "loss": 0.542,
+      "step": 3615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5362187708822924e-06,
+      "loss": 0.5339,
+      "step": 3616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531080300182965e-06,
+      "loss": 0.5231,
+      "step": 3617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525942523596334e-06,
+      "loss": 0.582,
+      "step": 3618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5208054425668405e-06,
+      "loss": 0.462,
+      "step": 3619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515669058538724e-06,
+      "loss": 0.5737,
+      "step": 3620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510533372956028e-06,
+      "loss": 0.6088,
+      "step": 3621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505398387262598e-06,
+      "loss": 0.6042,
+      "step": 3622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500264102902093e-06,
+      "loss": 0.5476,
+      "step": 3623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4951305213179615e-06,
+      "loss": 0.4985,
+      "step": 3624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489997643953455e-06,
+      "loss": 0.6833,
+      "step": 3625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484865472251639e-06,
+      "loss": 0.6636,
+      "step": 3626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4797340076553705e-06,
+      "loss": 0.5862,
+      "step": 3627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4746032516073154e-06,
+      "loss": 0.5898,
+      "step": 3628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469473205549927e-06,
+      "loss": 0.6514,
+      "step": 3629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464343870925479e-06,
+      "loss": 0.6436,
+      "step": 3630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459215249176021e-06,
+      "loss": 0.6196,
+      "step": 3631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4540873417434265e-06,
+      "loss": 0.7061,
+      "step": 3632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448960150069349e-06,
+      "loss": 0.7373,
+      "step": 3633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443833675595254e-06,
+      "loss": 0.5916,
+      "step": 3634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.438707919762395e-06,
+      "loss": 0.6125,
+      "step": 3635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433582884011833e-06,
+      "loss": 0.5474,
+      "step": 3636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428458569784418e-06,
+      "loss": 0.717,
+      "step": 3637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4233349785208e-06,
+      "loss": 0.6038,
+      "step": 3638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418212111661431e-06,
+      "loss": 0.6023,
+      "step": 3639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413089970646552e-06,
+      "loss": 0.5549,
+      "step": 3640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407968556916207e-06,
+      "loss": 0.6753,
+      "step": 3641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4028478719102235e-06,
+      "loss": 0.6528,
+      "step": 3642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39772791706824e-06,
+      "loss": 0.543,
+      "step": 3643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392608693829678e-06,
+      "loss": 0.4939,
+      "step": 3644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3874902036337595e-06,
+      "loss": 0.6331,
+      "step": 3645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3823724479194944e-06,
+      "loss": 0.5073,
+      "step": 3646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377255428125696e-06,
+      "loss": 0.5802,
+      "step": 3647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37213914569096e-06,
+      "loss": 0.6509,
+      "step": 3648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367023602053679e-06,
+      "loss": 0.6565,
+      "step": 3649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36190879865204e-06,
+      "loss": 0.5278,
+      "step": 3650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3567947369240155e-06,
+      "loss": 0.585,
+      "step": 3651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351681418307384e-06,
+      "loss": 0.5815,
+      "step": 3652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346568844239694e-06,
+      "loss": 0.6707,
+      "step": 3653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341457016158307e-06,
+      "loss": 0.5588,
+      "step": 3654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336345935500353e-06,
+      "loss": 0.6282,
+      "step": 3655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331235603702769e-06,
+      "loss": 0.6069,
+      "step": 3656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326126022202269e-06,
+      "loss": 0.5818,
+      "step": 3657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321017192435374e-06,
+      "loss": 0.5996,
+      "step": 3658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315909115838367e-06,
+      "loss": 0.593,
+      "step": 3659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 0.7141,
+      "step": 3660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305695227898175e-06,
+      "loss": 0.6516,
+      "step": 3661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30058941942652e-06,
+      "loss": 0.5779,
+      "step": 3662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295484369867834e-06,
+      "loss": 0.7351,
+      "step": 3663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29038008065734e-06,
+      "loss": 0.6208,
+      "step": 3664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285276553230071e-06,
+      "loss": 0.5698,
+      "step": 3665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2801737890208276e-06,
+      "loss": 0.5425,
+      "step": 3666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275071789464203e-06,
+      "loss": 0.7087,
+      "step": 3667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269970555994574e-06,
+      "loss": 0.4851,
+      "step": 3668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264870090046108e-06,
+      "loss": 0.6816,
+      "step": 3669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259770393052743e-06,
+      "loss": 0.5194,
+      "step": 3670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2546714664482175e-06,
+      "loss": 0.5187,
+      "step": 3671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249573311666038e-06,
+      "loss": 0.6714,
+      "step": 3672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244475930139505e-06,
+      "loss": 0.5947,
+      "step": 3673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239379323301696e-06,
+      "loss": 0.7068,
+      "step": 3674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23428349258547e-06,
+      "loss": 0.6206,
+      "step": 3675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229188439423478e-06,
+      "loss": 0.7061,
+      "step": 3676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224094165248135e-06,
+      "loss": 0.5916,
+      "step": 3677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219000671491654e-06,
+      "loss": 0.6394,
+      "step": 3678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213907959586015e-06,
+      "loss": 0.644,
+      "step": 3679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208816030962989e-06,
+      "loss": 0.6021,
+      "step": 3680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203724887054117e-06,
+      "loss": 0.5857,
+      "step": 3681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198634529290733e-06,
+      "loss": 0.7188,
+      "step": 3682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1935449591039334e-06,
+      "loss": 0.572,
+      "step": 3683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1884561779246055e-06,
+      "loss": 0.7271,
+      "step": 3684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1833681871834085e-06,
+      "loss": 0.6025,
+      "step": 3685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178280988310779e-06,
+      "loss": 0.646,
+      "step": 3686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173194582736942e-06,
+      "loss": 0.6782,
+      "step": 3687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1681089718918804e-06,
+      "loss": 0.6938,
+      "step": 3688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1630241572053734e-06,
+      "loss": 0.6467,
+      "step": 3689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157940140106958e-06,
+      "loss": 0.6265,
+      "step": 3690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152856922025962e-06,
+      "loss": 0.6226,
+      "step": 3691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14777450439148e-06,
+      "loss": 0.5898,
+      "step": 3692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1426928886323896e-06,
+      "loss": 0.645,
+      "step": 3693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137612076177329e-06,
+      "loss": 0.5212,
+      "step": 3694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132532068454727e-06,
+      "loss": 0.6194,
+      "step": 3695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.127452866892773e-06,
+      "loss": 0.6316,
+      "step": 3696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122374472919434e-06,
+      "loss": 0.468,
+      "step": 3697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117296887962458e-06,
+      "loss": 0.6528,
+      "step": 3698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112220113449351e-06,
+      "loss": 0.6694,
+      "step": 3699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107144150807403e-06,
+      "loss": 0.5762,
+      "step": 3700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102069001463668e-06,
+      "loss": 0.615,
+      "step": 3701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0969946668449805e-06,
+      "loss": 0.4856,
+      "step": 3702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091921148377932e-06,
+      "loss": 0.4821,
+      "step": 3703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086848447488901e-06,
+      "loss": 0.5776,
+      "step": 3704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08177656560402e-06,
+      "loss": 0.7314,
+      "step": 3705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076705504149208e-06,
+      "loss": 0.625,
+      "step": 3706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071635264550136e-06,
+      "loss": 0.4934,
+      "step": 3707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066565848232261e-06,
+      "loss": 0.5854,
+      "step": 3708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061497256620793e-06,
+      "loss": 0.5701,
+      "step": 3709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056429491140717e-06,
+      "loss": 0.5515,
+      "step": 3710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051362553216792e-06,
+      "loss": 0.6497,
+      "step": 3711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046296444273531e-06,
+      "loss": 0.507,
+      "step": 3712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041231165735227e-06,
+      "loss": 0.5287,
+      "step": 3713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036166719025927e-06,
+      "loss": 0.6802,
+      "step": 3714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031103105569459e-06,
+      "loss": 0.5742,
+      "step": 3715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026040326789402e-06,
+      "loss": 0.5818,
+      "step": 3716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020978384109113e-06,
+      "loss": 0.7041,
+      "step": 3717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015917278951699e-06,
+      "loss": 0.5686,
+      "step": 3718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010857012740051e-06,
+      "loss": 0.5833,
+      "step": 3719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0057975868968066e-06,
+      "loss": 0.6003,
+      "step": 3720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000739002844372e-06,
+      "loss": 0.6997,
+      "step": 3721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995681262004928e-06,
+      "loss": 0.5923,
+      "step": 3722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990624365800398e-06,
+      "loss": 0.6431,
+      "step": 3723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985568315652489e-06,
+      "loss": 0.5273,
+      "step": 3724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980513112982652e-06,
+      "loss": 0.6299,
+      "step": 3725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975458759212113e-06,
+      "loss": 0.6401,
+      "step": 3726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970405255761854e-06,
+      "loss": 0.6316,
+      "step": 3727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965352604052615e-06,
+      "loss": 0.6499,
+      "step": 3728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960300805504902e-06,
+      "loss": 0.644,
+      "step": 3729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9552498615389826e-06,
+      "loss": 0.6885,
+      "step": 3730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950199773574873e-06,
+      "loss": 0.553,
+      "step": 3731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945150543032362e-06,
+      "loss": 0.5452,
+      "step": 3732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940102171330989e-06,
+      "loss": 0.5061,
+      "step": 3733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935054659890053e-06,
+      "loss": 0.5471,
+      "step": 3734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9300080101286175e-06,
+      "loss": 0.5657,
+      "step": 3735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9249622234654925e-06,
+      "loss": 0.5889,
+      "step": 3736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919917301319259e-06,
+      "loss": 0.6646,
+      "step": 3737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914873245108239e-06,
+      "loss": 0.5593,
+      "step": 3738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4475,
+      "step": 3739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904787736163963e-06,
+      "loss": 0.6982,
+      "step": 3740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899746286266149e-06,
+      "loss": 0.7131,
+      "step": 3741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894705707974434e-06,
+      "loss": 0.5361,
+      "step": 3742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889666002705935e-06,
+      "loss": 0.6245,
+      "step": 3743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8846271718775106e-06,
+      "loss": 0.7009,
+      "step": 3744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879589216905777e-06,
+      "loss": 0.4954,
+      "step": 3745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874552139207116e-06,
+      "loss": 0.6321,
+      "step": 3746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86951594019764e-06,
+      "loss": 0.5828,
+      "step": 3747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864480621293237e-06,
+      "loss": 0.5383,
+      "step": 3748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8594461839095315e-06,
+      "loss": 0.6316,
+      "step": 3749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.854412629461911e-06,
+      "loss": 0.6245,
+      "step": 3750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849379959365509e-06,
+      "loss": 0.6614,
+      "step": 3751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84434817503521e-06,
+      "loss": 0.4802,
+      "step": 3752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839317277885651e-06,
+      "loss": 0.5818,
+      "step": 3753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834287269331226e-06,
+      "loss": 0.6567,
+      "step": 3754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829258150786062e-06,
+      "loss": 0.5906,
+      "step": 3755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824229923664058e-06,
+      "loss": 0.6528,
+      "step": 3756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819202589378841e-06,
+      "loss": 0.5435,
+      "step": 3757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814176149343802e-06,
+      "loss": 0.6404,
+      "step": 3758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809150604972079e-06,
+      "loss": 0.6143,
+      "step": 3759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804125957676546e-06,
+      "loss": 0.6174,
+      "step": 3760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799102208869842e-06,
+      "loss": 0.6602,
+      "step": 3761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794079359964339e-06,
+      "loss": 0.656,
+      "step": 3762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789057412372165e-06,
+      "loss": 0.5427,
+      "step": 3763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784036367505189e-06,
+      "loss": 0.5305,
+      "step": 3764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779016226775037e-06,
+      "loss": 0.5471,
+      "step": 3765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773996991593061e-06,
+      "loss": 0.5198,
+      "step": 3766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768978663370379e-06,
+      "loss": 0.4783,
+      "step": 3767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7639612435178404e-06,
+      "loss": 0.5812,
+      "step": 3768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758944733446043e-06,
+      "loss": 0.6995,
+      "step": 3769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753929134565336e-06,
+      "loss": 0.6565,
+      "step": 3770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748914448285797e-06,
+      "loss": 0.6003,
+      "step": 3771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743900676017265e-06,
+      "loss": 0.6514,
+      "step": 3772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738887819169307e-06,
+      "loss": 0.4976,
+      "step": 3773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733875879151241e-06,
+      "loss": 0.6792,
+      "step": 3774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7288648573721236e-06,
+      "loss": 0.7161,
+      "step": 3775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723854755240761e-06,
+      "loss": 0.5234,
+      "step": 3776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718845574165684e-06,
+      "loss": 0.535,
+      "step": 3777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713837315555187e-06,
+      "loss": 0.6628,
+      "step": 3778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708829980817282e-06,
+      "loss": 0.5145,
+      "step": 3779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703823571359741e-06,
+      "loss": 0.6797,
+      "step": 3780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698818088590062e-06,
+      "loss": 0.6016,
+      "step": 3781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693813533915485e-06,
+      "loss": 0.7373,
+      "step": 3782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688809908743002e-06,
+      "loss": 0.5479,
+      "step": 3783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683807214479323e-06,
+      "loss": 0.6243,
+      "step": 3784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6788054525309145e-06,
+      "loss": 0.6116,
+      "step": 3785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673804624303966e-06,
+      "loss": 0.4606,
+      "step": 3786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.668804731204417e-06,
+      "loss": 0.7261,
+      "step": 3787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663805774637935e-06,
+      "loss": 0.8311,
+      "step": 3788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658807756009933e-06,
+      "loss": 0.7053,
+      "step": 3789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653810676725548e-06,
+      "loss": 0.5671,
+      "step": 3790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648814538189668e-06,
+      "loss": 0.7112,
+      "step": 3791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643819341806902e-06,
+      "loss": 0.6541,
+      "step": 3792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638825088981604e-06,
+      "loss": 0.6365,
+      "step": 3793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633831781117862e-06,
+      "loss": 0.6426,
+      "step": 3794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6288394196194885e-06,
+      "loss": 0.666,
+      "step": 3795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623848005890046e-06,
+      "loss": 0.5403,
+      "step": 3796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618857541332812e-06,
+      "loss": 0.5137,
+      "step": 3797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613868027350816e-06,
+      "loss": 0.6616,
+      "step": 3798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608879465346804e-06,
+      "loss": 0.6284,
+      "step": 3799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60389185672327e-06,
+      "loss": 0.5143,
+      "step": 3800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598905202882422e-06,
+      "loss": 0.7109,
+      "step": 3801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593919505226218e-06,
+      "loss": 0.5819,
+      "step": 3802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58893476515633e-06,
+      "loss": 0.6375,
+      "step": 3803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583950984074175e-06,
+      "loss": 0.5496,
+      "step": 3804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5789681633808945e-06,
+      "loss": 0.7368,
+      "step": 3805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573986304477353e-06,
+      "loss": 0.6436,
+      "step": 3806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569005408764162e-06,
+      "loss": 0.6021,
+      "step": 3807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56402547764164e-06,
+      "loss": 0.6821,
+      "step": 3808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559046512509859e-06,
+      "loss": 0.6233,
+      "step": 3809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554068514768595e-06,
+      "loss": 0.6514,
+      "step": 3810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549091485817369e-06,
+      "loss": 0.5986,
+      "step": 3811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544115427055424e-06,
+      "loss": 0.5654,
+      "step": 3812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.539140339881732e-06,
+      "loss": 0.4201,
+      "step": 3813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5341662256949844e-06,
+      "loss": 0.5994,
+      "step": 3814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529193085893613e-06,
+      "loss": 0.6572,
+      "step": 3815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524220921875759e-06,
+      "loss": 0.5872,
+      "step": 3816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519249735039306e-06,
+      "loss": 0.4905,
+      "step": 3817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 0.5474,
+      "step": 3818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5093102985007154e-06,
+      "loss": 0.6763,
+      "step": 3819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504342051592958e-06,
+      "loss": 0.7327,
+      "step": 3820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499374787455344e-06,
+      "loss": 0.7202,
+      "step": 3821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4944085074843754e-06,
+      "loss": 0.5498,
+      "step": 3822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4894432130762745e-06,
+      "loss": 0.6643,
+      "step": 3823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484478905626988e-06,
+      "loss": 0.5543,
+      "step": 3824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479515586532172e-06,
+      "loss": 0.7031,
+      "step": 3825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474553257187229e-06,
+      "loss": 0.8035,
+      "step": 3826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469591918987258e-06,
+      "loss": 0.5916,
+      "step": 3827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464631573327098e-06,
+      "loss": 0.6443,
+      "step": 3828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459672221601301e-06,
+      "loss": 0.6636,
+      "step": 3829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454713865204134e-06,
+      "loss": 0.614,
+      "step": 3830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449756505529601e-06,
+      "loss": 0.7234,
+      "step": 3831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444800143971404e-06,
+      "loss": 0.6609,
+      "step": 3832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439844781922986e-06,
+      "loss": 0.6902,
+      "step": 3833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434890420777491e-06,
+      "loss": 0.6714,
+      "step": 3834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429937061927795e-06,
+      "loss": 0.7192,
+      "step": 3835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42498470676648e-06,
+      "loss": 0.697,
+      "step": 3836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420033356685862e-06,
+      "loss": 0.6169,
+      "step": 3837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415083013077954e-06,
+      "loss": 0.646,
+      "step": 3838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410133677334506e-06,
+      "loss": 0.6716,
+      "step": 3839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405185350846967e-06,
+      "loss": 0.5027,
+      "step": 3840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400238035006519e-06,
+      "loss": 0.6416,
+      "step": 3841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395291731204049e-06,
+      "loss": 0.731,
+      "step": 3842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390346440830156e-06,
+      "loss": 0.5544,
+      "step": 3843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38540216527517e-06,
+      "loss": 0.5693,
+      "step": 3844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380458905929117e-06,
+      "loss": 0.5444,
+      "step": 3845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375516664181751e-06,
+      "loss": 0.613,
+      "step": 3846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370575441422533e-06,
+      "loss": 0.6404,
+      "step": 3847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365635239040646e-06,
+      "loss": 0.6475,
+      "step": 3848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360696058424969e-06,
+      "loss": 0.6377,
+      "step": 3849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3557579009641144e-06,
+      "loss": 0.4897,
+      "step": 3850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350820768046389e-06,
+      "loss": 0.5696,
+      "step": 3851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345884661059825e-06,
+      "loss": 0.5171,
+      "step": 3852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340949581392159e-06,
+      "loss": 0.6753,
+      "step": 3853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336015530430839e-06,
+      "loss": 0.5728,
+      "step": 3854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331082509563029e-06,
+      "loss": 0.6074,
+      "step": 3855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326150520175596e-06,
+      "loss": 0.6501,
+      "step": 3856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321219563655121e-06,
+      "loss": 0.5359,
+      "step": 3857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316289641387896e-06,
+      "loss": 0.499,
+      "step": 3858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311360754759923e-06,
+      "loss": 0.4884,
+      "step": 3859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306432905156904e-06,
+      "loss": 0.6494,
+      "step": 3860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301506093964264e-06,
+      "loss": 0.6086,
+      "step": 3861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296580322567119e-06,
+      "loss": 0.6416,
+      "step": 3862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29165559235031e-06,
+      "loss": 0.5037,
+      "step": 3863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286731904698368e-06,
+      "loss": 0.5515,
+      "step": 3864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2818092609955465e-06,
+      "loss": 0.6523,
+      "step": 3865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276887662625799e-06,
+      "loss": 0.6797,
+      "step": 3866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27196711097278e-06,
+      "loss": 0.6753,
+      "step": 3867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2670476074198606e-06,
+      "loss": 0.5262,
+      "step": 3868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262129153350105e-06,
+      "loss": 0.7314,
+      "step": 3869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.257211750146292e-06,
+      "loss": 0.5698,
+      "step": 3870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252295399190899e-06,
+      "loss": 0.5511,
+      "step": 3871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2473801018661185e-06,
+      "loss": 0.6116,
+      "step": 3872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242465859553828e-06,
+      "loss": 0.5715,
+      "step": 3873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237552673635625e-06,
+      "loss": 0.592,
+      "step": 3874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2326405454928e-06,
+      "loss": 0.7808,
+      "step": 3875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227729476506352e-06,
+      "loss": 0.6116,
+      "step": 3876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222819468056984e-06,
+      "loss": 0.7212,
+      "step": 3877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217910521525088e-06,
+      "loss": 0.522,
+      "step": 3878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213002638290775e-06,
+      "loss": 0.6274,
+      "step": 3879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2080958197338415e-06,
+      "loss": 0.666,
+      "step": 3880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203190067233798e-06,
+      "loss": 0.7495,
+      "step": 3881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198285382169843e-06,
+      "loss": 0.5879,
+      "step": 3882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193381765920888e-06,
+      "loss": 0.6274,
+      "step": 3883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188479219865529e-06,
+      "loss": 0.6282,
+      "step": 3884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183577745382077e-06,
+      "loss": 0.6228,
+      "step": 3885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1786773438485274e-06,
+      "loss": 0.6138,
+      "step": 3886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173778016642582e-06,
+      "loss": 0.7007,
+      "step": 3887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16887976514164e-06,
+      "loss": 0.6201,
+      "step": 3888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1639825907227964e-06,
+      "loss": 0.6919,
+      "step": 3889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159086494762848e-06,
+      "loss": 0.562,
+      "step": 3890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154191478638276e-06,
+      "loss": 0.6582,
+      "step": 3891
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149297543725274e-06,
+      "loss": 0.5857,
+      "step": 3892
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144404691399717e-06,
+      "loss": 0.6147,
+      "step": 3893
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13951292303719e-06,
+      "loss": 0.5049,
+      "step": 3894
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1346222400129595e-06,
+      "loss": 0.5747,
+      "step": 3895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129732643702002e-06,
+      "loss": 0.4752,
+      "step": 3896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 0.5435,
+      "step": 3897
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11995671671823e-06,
+      "loss": 0.5349,
+      "step": 3898
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1150703887938205e-06,
+      "loss": 0.5472,
+      "step": 3899
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110185153079493e-06,
+      "loss": 0.5405,
+      "step": 3900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105301010948684e-06,
+      "loss": 0.5125,
+      "step": 3901
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100417963774518e-06,
+      "loss": 0.4927,
+      "step": 3902
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095536012929821e-06,
+      "loss": 0.5012,
+      "step": 3903
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0906551597871e-06,
+      "loss": 0.5552,
+      "step": 3904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085775405718563e-06,
+      "loss": 0.6243,
+      "step": 3905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080896752096104e-06,
+      "loss": 0.48,
+      "step": 3906
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076019200291314e-06,
+      "loss": 0.623,
+      "step": 3907
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0711427516754604e-06,
+      "loss": 0.616,
+      "step": 3908
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06626740761952e-06,
+      "loss": 0.613,
+      "step": 3909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061393169494134e-06,
+      "loss": 0.564,
+      "step": 3910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05652003866966e-06,
+      "loss": 0.6196,
+      "step": 3911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051648016516124e-06,
+      "loss": 0.5894,
+      "step": 3912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04677710440325e-06,
+      "loss": 0.8005,
+      "step": 3913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041907303700449e-06,
+      "loss": 0.6052,
+      "step": 3914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037038615776814e-06,
+      "loss": 0.6587,
+      "step": 3915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032171042001132e-06,
+      "loss": 0.6724,
+      "step": 3916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0273045837418705e-06,
+      "loss": 0.6233,
+      "step": 3917
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022439242367193e-06,
+      "loss": 0.6909,
+      "step": 3918
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017575019244935e-06,
+      "loss": 0.5815,
+      "step": 3919
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012711915742635e-06,
+      "loss": 0.5989,
+      "step": 3920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007849933227496e-06,
+      "loss": 0.614,
+      "step": 3921
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002989073066427e-06,
+      "loss": 0.6396,
+      "step": 3922
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998129336626004e-06,
+      "loss": 0.6416,
+      "step": 3923
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9932707252725e-06,
+      "loss": 0.5503,
+      "step": 3924
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988413240371864e-06,
+      "loss": 0.6345,
+      "step": 3925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983556883289727e-06,
+      "loss": 0.6328,
+      "step": 3926
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9787016553914146e-06,
+      "loss": 0.6475,
+      "step": 3927
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973847558041917e-06,
+      "loss": 0.6323,
+      "step": 3928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968994592605925e-06,
+      "loss": 0.7588,
+      "step": 3929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.964142760447795e-06,
+      "loss": 0.6655,
+      "step": 3930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959292062931581e-06,
+      "loss": 0.6909,
+      "step": 3931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954442501420999e-06,
+      "loss": 0.5398,
+      "step": 3932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949594077279467e-06,
+      "loss": 0.6772,
+      "step": 3933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944746791870062e-06,
+      "loss": 0.6406,
+      "step": 3934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939900646555557e-06,
+      "loss": 0.5354,
+      "step": 3935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9350556426983956e-06,
+      "loss": 0.5886,
+      "step": 3936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930211781660708e-06,
+      "loss": 0.7073,
+      "step": 3937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925369064804296e-06,
+      "loss": 0.7087,
+      "step": 3938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920527493490638e-06,
+      "loss": 0.6118,
+      "step": 3939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9156870690809e-06,
+      "loss": 0.6531,
+      "step": 3940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910847792935918e-06,
+      "loss": 0.6577,
+      "step": 3941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906009666416208e-06,
+      "loss": 0.5236,
+      "step": 3942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901172690881959e-06,
+      "loss": 0.6301,
+      "step": 3943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896336867693046e-06,
+      "loss": 0.6599,
+      "step": 3944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891502198209006e-06,
+      "loss": 0.6477,
+      "step": 3945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886668683789067e-06,
+      "loss": 0.4971,
+      "step": 3946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881836325792116e-06,
+      "loss": 0.4989,
+      "step": 3947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877005125576732e-06,
+      "loss": 0.8113,
+      "step": 3948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872175084501154e-06,
+      "loss": 0.606,
+      "step": 3949
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8673462039232995e-06,
+      "loss": 0.6794,
+      "step": 3950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86251848520077e-06,
+      "loss": 0.7402,
+      "step": 3951
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857691929690819e-06,
+      "loss": 0.7158,
+      "step": 3952
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.852866538750396e-06,
+      "loss": 0.6099,
+      "step": 3953
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848042313736107e-06,
+      "loss": 0.5916,
+      "step": 3954
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84321925600424e-06,
+      "loss": 0.7078,
+      "step": 3955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838397366910747e-06,
+      "loss": 0.5796,
+      "step": 3956
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833576647811258e-06,
+      "loss": 0.5728,
+      "step": 3957
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828757100061067e-06,
+      "loss": 0.6267,
+      "step": 3958
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823938725015148e-06,
+      "loss": 0.4512,
+      "step": 3959
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.819121524028134e-06,
+      "loss": 0.5806,
+      "step": 3960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814305498454342e-06,
+      "loss": 0.616,
+      "step": 3961
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8094906496477425e-06,
+      "loss": 0.6675,
+      "step": 3962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804676978961987e-06,
+      "loss": 0.6221,
+      "step": 3963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799864487750395e-06,
+      "loss": 0.6438,
+      "step": 3964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795053177365947e-06,
+      "loss": 0.5857,
+      "step": 3965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7902430491613e-06,
+      "loss": 0.4868,
+      "step": 3966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78543410448877e-06,
+      "loss": 0.6226,
+      "step": 3967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780626344700349e-06,
+      "loss": 0.6389,
+      "step": 3968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7758197711476875e-06,
+      "loss": 0.5864,
+      "step": 3969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771014385182112e-06,
+      "loss": 0.6296,
+      "step": 3970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766210188154604e-06,
+      "loss": 0.5769,
+      "step": 3971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761407181415825e-06,
+      "loss": 0.6279,
+      "step": 3972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7566053663160825e-06,
+      "loss": 0.6035,
+      "step": 3973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751804744205367e-06,
+      "loss": 0.6021,
+      "step": 3974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74700531643333e-06,
+      "loss": 0.6211,
+      "step": 3975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.7537,
+      "step": 3976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737410049302183e-06,
+      "loss": 0.6199,
+      "step": 3977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732614212640693e-06,
+      "loss": 0.624,
+      "step": 3978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72781957571311e-06,
+      "loss": 0.6174,
+      "step": 3979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723026139867393e-06,
+      "loss": 0.5747,
+      "step": 3980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71823390645118e-06,
+      "loss": 0.6007,
+      "step": 3981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71344287681175e-06,
+      "loss": 0.5505,
+      "step": 3982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708653052296062e-06,
+      "loss": 0.4788,
+      "step": 3983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703864434250721e-06,
+      "loss": 0.5391,
+      "step": 3984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699077024022011e-06,
+      "loss": 0.6602,
+      "step": 3985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694290822955853e-06,
+      "loss": 0.6704,
+      "step": 3986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689505832397848e-06,
+      "loss": 0.6597,
+      "step": 3987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684722053693253e-06,
+      "loss": 0.5625,
+      "step": 3988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679939488186972e-06,
+      "loss": 0.6431,
+      "step": 3989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675158137223584e-06,
+      "loss": 0.5195,
+      "step": 3990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670378002147313e-06,
+      "loss": 0.6614,
+      "step": 3991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6655990843020546e-06,
+      "loss": 0.6748,
+      "step": 3992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660821385031346e-06,
+      "loss": 0.5476,
+      "step": 3993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656044905678402e-06,
+      "loss": 0.6206,
+      "step": 3994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651269647586071e-06,
+      "loss": 0.6199,
+      "step": 3995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646495612096881e-06,
+      "loss": 0.5356,
+      "step": 3996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641722800552994e-06,
+      "loss": 0.6455,
+      "step": 3997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636951214296249e-06,
+      "loss": 0.5745,
+      "step": 3998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63218085466813e-06,
+      "loss": 0.7197,
+      "step": 3999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627411723009772e-06,
+      "loss": 0.5684,
+      "step": 4000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622643820661979e-06,
+      "loss": 0.5532,
+      "step": 4001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617877148965187e-06,
+      "loss": 0.5908,
+      "step": 4002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613111709259512e-06,
+      "loss": 0.8196,
+      "step": 4003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608347502884701e-06,
+      "loss": 0.6125,
+      "step": 4004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603584531180173e-06,
+      "loss": 0.688,
+      "step": 4005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5988227954849826e-06,
+      "loss": 0.6121,
+      "step": 4006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594062297137853e-06,
+      "loss": 0.6343,
+      "step": 4007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589303037477145e-06,
+      "loss": 0.5896,
+      "step": 4008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584545017840886e-06,
+      "loss": 0.5852,
+      "step": 4009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579788239566736e-06,
+      "loss": 0.7451,
+      "step": 4010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5750327039920235e-06,
+      "loss": 0.542,
+      "step": 4011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570278412453725e-06,
+      "loss": 0.585,
+      "step": 4012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565525366288455e-06,
+      "loss": 0.5852,
+      "step": 4013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560773566832494e-06,
+      "loss": 0.4613,
+      "step": 4014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.556023015421755e-06,
+      "loss": 0.7927,
+      "step": 4015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5512737133918175e-06,
+      "loss": 0.584,
+      "step": 4016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546525662077895e-06,
+      "loss": 0.6147,
+      "step": 4017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541778862814862e-06,
+      "loss": 0.7393,
+      "step": 4018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537033316937228e-06,
+      "loss": 0.481,
+      "step": 4019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532289025779165e-06,
+      "loss": 0.6548,
+      "step": 4020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527545990674475e-06,
+      "loss": 0.6743,
+      "step": 4021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522804212956621e-06,
+      "loss": 0.5408,
+      "step": 4022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518063693958713e-06,
+      "loss": 0.5645,
+      "step": 4023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51332443501349e-06,
+      "loss": 0.5481,
+      "step": 4024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508586437453359e-06,
+      "loss": 0.6628,
+      "step": 4025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503849702610354e-06,
+      "loss": 0.5764,
+      "step": 4026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49911423181617e-06,
+      "loss": 0.6677,
+      "step": 4027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4943800264021285e-06,
+      "loss": 0.6538,
+      "step": 4028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489647087699215e-06,
+      "loss": 0.6653,
+      "step": 4029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4849154170380395e-06,
+      "loss": 0.5696,
+      "step": 4030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4801850157488746e-06,
+      "loss": 0.6716,
+      "step": 4031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475455885161618e-06,
+      "loss": 0.5525,
+      "step": 4032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4707280266058215e-06,
+      "loss": 0.592,
+      "step": 4033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4660014414106825e-06,
+      "loss": 0.6709,
+      "step": 4034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461276130905024e-06,
+      "loss": 0.6833,
+      "step": 4035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4565520964173326e-06,
+      "loss": 0.6976,
+      "step": 4036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4518293392757125e-06,
+      "loss": 0.6323,
+      "step": 4037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4471078608079306e-06,
+      "loss": 0.6941,
+      "step": 4038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442387662341377e-06,
+      "loss": 0.6465,
+      "step": 4039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437668745203097e-06,
+      "loss": 0.7058,
+      "step": 4040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43295111071976e-06,
+      "loss": 0.5786,
+      "step": 4041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42823476021769e-06,
+      "loss": 0.5662,
+      "step": 4042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4235196950228375e-06,
+      "loss": 0.6641,
+      "step": 4043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418805916460803e-06,
+      "loss": 0.5247,
+      "step": 4044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4140934258568125e-06,
+      "loss": 0.6948,
+      "step": 4045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409382224535738e-06,
+      "loss": 0.6611,
+      "step": 4046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404672313822097e-06,
+      "loss": 0.5472,
+      "step": 4047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399963695040022e-06,
+      "loss": 0.5488,
+      "step": 4048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3952563695133065e-06,
+      "loss": 0.6597,
+      "step": 4049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390550338565359e-06,
+      "loss": 0.5991,
+      "step": 4050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3858456035192444e-06,
+      "loss": 0.6646,
+      "step": 4051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381142165697641e-06,
+      "loss": 0.6892,
+      "step": 4052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376440026422889e-06,
+      "loss": 0.6255,
+      "step": 4053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371739187016935e-06,
+      "loss": 0.699,
+      "step": 4054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 0.6211,
+      "step": 4055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36234141309746e-06,
+      "loss": 0.613,
+      "step": 4056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357644481226027e-06,
+      "loss": 0.5787,
+      "step": 4057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352948854507588e-06,
+      "loss": 0.6816,
+      "step": 4058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348254534262262e-06,
+      "loss": 0.5759,
+      "step": 4059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3435615218098216e-06,
+      "loss": 0.6248,
+      "step": 4060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338869818469651e-06,
+      "loss": 0.5852,
+      "step": 4061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334179425560786e-06,
+      "loss": 0.5767,
+      "step": 4062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32949034440188e-06,
+      "loss": 0.5393,
+      "step": 4063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324802576311226e-06,
+      "loss": 0.603,
+      "step": 4064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320116122606736e-06,
+      "loss": 0.5555,
+      "step": 4065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315430984605974e-06,
+      "loss": 0.6814,
+      "step": 4066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310747163626108e-06,
+      "loss": 0.6812,
+      "step": 4067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306064660983958e-06,
+      "loss": 0.6504,
+      "step": 4068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3013834779959564e-06,
+      "loss": 0.751,
+      "step": 4069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296703615978176e-06,
+      "loss": 0.5962,
+      "step": 4070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292025076246319e-06,
+      "loss": 0.6736,
+      "step": 4071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287347860115701e-06,
+      "loss": 0.5928,
+      "step": 4072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282671968901287e-06,
+      "loss": 0.7473,
+      "step": 4073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277997403917648e-06,
+      "loss": 0.6218,
+      "step": 4074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2733241664789994e-06,
+      "loss": 0.6084,
+      "step": 4075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26865225789917e-06,
+      "loss": 0.6482,
+      "step": 4076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26398167949163e-06,
+      "loss": 0.6665,
+      "step": 4077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259312432569458e-06,
+      "loss": 0.5669,
+      "step": 4078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254644518445375e-06,
+      "loss": 0.5544,
+      "step": 4079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2499779384317115e-06,
+      "loss": 0.6812,
+      "step": 4080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245312693840436e-06,
+      "loss": 0.4935,
+      "step": 4081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24064878598314e-06,
+      "loss": 0.6599,
+      "step": 4082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235986216171027e-06,
+      "loss": 0.5867,
+      "step": 4083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231324985714942e-06,
+      "loss": 0.5151,
+      "step": 4084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2266650959253365e-06,
+      "loss": 0.6575,
+      "step": 4085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222006548112299e-06,
+      "loss": 0.6697,
+      "step": 4086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217349343585529e-06,
+      "loss": 0.5823,
+      "step": 4087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212693483654363e-06,
+      "loss": 0.6172,
+      "step": 4088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20803896962774e-06,
+      "loss": 0.679,
+      "step": 4089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20338580281424e-06,
+      "loss": 0.6292,
+      "step": 4090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19873398452205e-06,
+      "loss": 0.5898,
+      "step": 4091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194083516058987e-06,
+      "loss": 0.6384,
+      "step": 4092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1894343987324805e-06,
+      "loss": 0.5903,
+      "step": 4093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184786633849588e-06,
+      "loss": 0.6389,
+      "step": 4094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180140222716985e-06,
+      "loss": 0.613,
+      "step": 4095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175495166640958e-06,
+      "loss": 0.5349,
+      "step": 4096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170851466927428e-06,
+      "loss": 0.6543,
+      "step": 4097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166209124881918e-06,
+      "loss": 0.6355,
+      "step": 4098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161568141809583e-06,
+      "loss": 0.5974,
+      "step": 4099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156928519015184e-06,
+      "loss": 0.7124,
+      "step": 4100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152290257803115e-06,
+      "loss": 0.7114,
+      "step": 4101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147653359477365e-06,
+      "loss": 0.5764,
+      "step": 4102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1430178253415655e-06,
+      "loss": 0.5671,
+      "step": 4103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138383656698942e-06,
+      "loss": 0.6011,
+      "step": 4104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13375085485235e-06,
+      "loss": 0.7153,
+      "step": 4105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129119421104263e-06,
+      "loss": 0.686,
+      "step": 4106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12448935675675e-06,
+      "loss": 0.5969,
+      "step": 4107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11986066311152e-06,
+      "loss": 0.6257,
+      "step": 4108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1152333414698774e-06,
+      "loss": 0.5006,
+      "step": 4109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.110607393132756e-06,
+      "loss": 0.6118,
+      "step": 4110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105982819400687e-06,
+      "loss": 0.5952,
+      "step": 4111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1013596215738326e-06,
+      "loss": 0.4958,
+      "step": 4112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096737800951953e-06,
+      "loss": 0.7166,
+      "step": 4113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092117358834434e-06,
+      "loss": 0.5334,
+      "step": 4114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0874982965202635e-06,
+      "loss": 0.6211,
+      "step": 4115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082880615308045e-06,
+      "loss": 0.6497,
+      "step": 4116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078264316496002e-06,
+      "loss": 0.6931,
+      "step": 4117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0736494013819525e-06,
+      "loss": 0.6218,
+      "step": 4118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069035871263344e-06,
+      "loss": 0.5735,
+      "step": 4119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064423727437215e-06,
+      "loss": 0.6973,
+      "step": 4120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059812971200235e-06,
+      "loss": 0.5825,
+      "step": 4121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.055203603848666e-06,
+      "loss": 0.6089,
+      "step": 4122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050595626678392e-06,
+      "loss": 0.5635,
+      "step": 4123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045989040984896e-06,
+      "loss": 0.5967,
+      "step": 4124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.04138384806328e-06,
+      "loss": 0.6233,
+      "step": 4125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036780049208239e-06,
+      "loss": 0.5818,
+      "step": 4126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0321776457141004e-06,
+      "loss": 0.5916,
+      "step": 4127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0275766388747725e-06,
+      "loss": 0.5718,
+      "step": 4128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0229770299837886e-06,
+      "loss": 0.7628,
+      "step": 4129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018378820334289e-06,
+      "loss": 0.5537,
+      "step": 4130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.013782011219006e-06,
+      "loss": 0.6665,
+      "step": 4131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009186603930296e-06,
+      "loss": 0.7695,
+      "step": 4132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0045925997601065e-06,
+      "loss": 0.6499,
+      "step": 4133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5686,
+      "step": 4134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995408805941142e-06,
+      "loss": 0.5391,
+      "step": 4135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990819018874302e-06,
+      "loss": 0.6943,
+      "step": 4136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986230640089849e-06,
+      "loss": 0.6208,
+      "step": 4137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9816436708777685e-06,
+      "loss": 0.7441,
+      "step": 4138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9770581125276345e-06,
+      "loss": 0.6831,
+      "step": 4139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9724739663286344e-06,
+      "loss": 0.6418,
+      "step": 4140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967891233569563e-06,
+      "loss": 0.8213,
+      "step": 4141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963309915538801e-06,
+      "loss": 0.627,
+      "step": 4142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958730013524349e-06,
+      "loss": 0.5583,
+      "step": 4143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954151528813796e-06,
+      "loss": 0.709,
+      "step": 4144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949574462694344e-06,
+      "loss": 0.5593,
+      "step": 4145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944998816452784e-06,
+      "loss": 0.625,
+      "step": 4146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940424591375521e-06,
+      "loss": 0.5405,
+      "step": 4147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935851788748547e-06,
+      "loss": 0.676,
+      "step": 4148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9312804098574685e-06,
+      "loss": 0.5498,
+      "step": 4149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926710455987475e-06,
+      "loss": 0.5649,
+      "step": 4150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922141928423375e-06,
+      "loss": 0.6499,
+      "step": 4151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917574828449555e-06,
+      "loss": 0.5867,
+      "step": 4152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913009157350016e-06,
+      "loss": 0.7822,
+      "step": 4153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908444916408351e-06,
+      "loss": 0.6719,
+      "step": 4154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903882106907757e-06,
+      "loss": 0.5367,
+      "step": 4155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8993207301310185e-06,
+      "loss": 0.5881,
+      "step": 4156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894760787360519e-06,
+      "loss": 0.6084,
+      "step": 4157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8902022798782475e-06,
+      "loss": 0.5679,
+      "step": 4158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885645208965779e-06,
+      "loss": 0.7212,
+      "step": 4159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8810895759042955e-06,
+      "loss": 0.6748,
+      "step": 4160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876535381974562e-06,
+      "loss": 0.5608,
+      "step": 4161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871982628456954e-06,
+      "loss": 0.7544,
+      "step": 4162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867431316631425e-06,
+      "loss": 0.6545,
+      "step": 4163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8628814477775345e-06,
+      "loss": 0.6604,
+      "step": 4164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858333023174436e-06,
+      "loss": 0.5549,
+      "step": 4165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853786044100879e-06,
+      "loss": 0.7188,
+      "step": 4166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849240511835196e-06,
+      "loss": 0.4822,
+      "step": 4167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844696427655317e-06,
+      "loss": 0.6357,
+      "step": 4168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840153792838774e-06,
+      "loss": 0.729,
+      "step": 4169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835612608662678e-06,
+      "loss": 0.6394,
+      "step": 4170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831072876403744e-06,
+      "loss": 0.635,
+      "step": 4171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826534597338267e-06,
+      "loss": 0.5498,
+      "step": 4172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821997772742149e-06,
+      "loss": 0.7231,
+      "step": 4173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8174624038908645e-06,
+      "loss": 0.7275,
+      "step": 4174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812928492059496e-06,
+      "loss": 0.5664,
+      "step": 4175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808396038522701e-06,
+      "loss": 0.5278,
+      "step": 4176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803865044554739e-06,
+      "loss": 0.6077,
+      "step": 4177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7993355114294556e-06,
+      "loss": 0.5693,
+      "step": 4178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7948074404202886e-06,
+      "loss": 0.6008,
+      "step": 4179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7902808328002535e-06,
+      "loss": 0.7349,
+      "step": 4180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785755689841962e-06,
+      "loss": 0.5164,
+      "step": 4181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781232012817621e-06,
+      "loss": 0.564,
+      "step": 4182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77670980299901e-06,
+      "loss": 0.6145,
+      "step": 4183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772189061657511e-06,
+      "loss": 0.6736,
+      "step": 4184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76766979006408e-06,
+      "loss": 0.6992,
+      "step": 4185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763151989489273e-06,
+      "loss": 0.689,
+      "step": 4186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758635661203218e-06,
+      "loss": 0.6589,
+      "step": 4187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754120806475641e-06,
+      "loss": 0.635,
+      "step": 4188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749607426575849e-06,
+      "loss": 0.71,
+      "step": 4189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745095522772739e-06,
+      "loss": 0.6453,
+      "step": 4190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740585096334784e-06,
+      "loss": 0.6816,
+      "step": 4191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736076148530044e-06,
+      "loss": 0.688,
+      "step": 4192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731568680626173e-06,
+      "loss": 0.7258,
+      "step": 4193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727062693890392e-06,
+      "loss": 0.6628,
+      "step": 4194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722558189589526e-06,
+      "loss": 0.7461,
+      "step": 4195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718055168989961e-06,
+      "loss": 0.6628,
+      "step": 4196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713553633357689e-06,
+      "loss": 0.6738,
+      "step": 4197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709053583958263e-06,
+      "loss": 0.6582,
+      "step": 4198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704555022056837e-06,
+      "loss": 0.6417,
+      "step": 4199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700057948918127e-06,
+      "loss": 0.709,
+      "step": 4200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69556236580645e-06,
+      "loss": 0.6184,
+      "step": 4201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691068273985692e-06,
+      "loss": 0.7212,
+      "step": 4202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686575674719327e-06,
+      "loss": 0.6335,
+      "step": 4203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682084569270402e-06,
+      "loss": 0.6221,
+      "step": 4204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677594958901544e-06,
+      "loss": 0.6116,
+      "step": 4205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6731068448749715e-06,
+      "loss": 0.6899,
+      "step": 4206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668620228452465e-06,
+      "loss": 0.6855,
+      "step": 4207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6641351108954015e-06,
+      "loss": 0.5686,
+      "step": 4208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659651493464721e-06,
+      "loss": 0.6172,
+      "step": 4209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6551693774209535e-06,
+      "loss": 0.5886,
+      "step": 4210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650688764024199e-06,
+      "loss": 0.6199,
+      "step": 4211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646209654534138e-06,
+      "loss": 0.5842,
+      "step": 4212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.6299,
+      "step": 4213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637255952310716e-06,
+      "loss": 0.4629,
+      "step": 4214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632781362094602e-06,
+      "loss": 0.6001,
+      "step": 4215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6283082808196685e-06,
+      "loss": 0.7004,
+      "step": 4216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62383670974349e-06,
+      "loss": 0.6138,
+      "step": 4217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619366650123196e-06,
+      "loss": 0.5955,
+      "step": 4218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614898103215507e-06,
+      "loss": 0.6736,
+      "step": 4219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610431070276706e-06,
+      "loss": 0.7354,
+      "step": 4220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605965552562661e-06,
+      "loss": 0.6482,
+      "step": 4221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601501551328804e-06,
+      "loss": 0.5159,
+      "step": 4222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597039067830147e-06,
+      "loss": 0.7544,
+      "step": 4223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592578103321277e-06,
+      "loss": 0.7222,
+      "step": 4224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588118659056346e-06,
+      "loss": 0.7256,
+      "step": 4225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583660736289084e-06,
+      "loss": 0.6467,
+      "step": 4226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579204336272797e-06,
+      "loss": 0.5266,
+      "step": 4227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574749460260356e-06,
+      "loss": 0.6067,
+      "step": 4228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570296109504201e-06,
+      "loss": 0.627,
+      "step": 4229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565844285256357e-06,
+      "loss": 0.6263,
+      "step": 4230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561393988768401e-06,
+      "loss": 0.543,
+      "step": 4231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5569452212915e-06,
+      "loss": 0.5518,
+      "step": 4232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552497984076374e-06,
+      "loss": 0.6433,
+      "step": 4233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548052278373327e-06,
+      "loss": 0.603,
+      "step": 4234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543608105432219e-06,
+      "loss": 0.5793,
+      "step": 4235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539165466502488e-06,
+      "loss": 0.6953,
+      "step": 4236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5347243628331425e-06,
+      "loss": 0.5825,
+      "step": 4237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5302847956727556e-06,
+      "loss": 0.5217,
+      "step": 4238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525846766269465e-06,
+      "loss": 0.5994,
+      "step": 4239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5214102758709774e-06,
+      "loss": 0.4258,
+      "step": 4240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516975325724575e-06,
+      "loss": 0.5073,
+      "step": 4241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5125419170770935e-06,
+      "loss": 0.7587,
+      "step": 4242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50811005117495e-06,
+      "loss": 0.7578,
+      "step": 4243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503679729264112e-06,
+      "loss": 0.5991,
+      "step": 4244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499250952590131e-06,
+      "loss": 0.6655,
+      "step": 4245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494823722398105e-06,
+      "loss": 0.5098,
+      "step": 4246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490398039932712e-06,
+      "loss": 0.575,
+      "step": 4247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485973906438191e-06,
+      "loss": 0.5962,
+      "step": 4248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4815513231583385e-06,
+      "loss": 0.6836,
+      "step": 4249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477130291336523e-06,
+      "loss": 0.7236,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472710812215679e-06,
+      "loss": 0.6038,
+      "step": 4251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468292887038297e-06,
+      "loss": 0.6858,
+      "step": 4252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463876517046427e-06,
+      "loss": 0.6067,
+      "step": 4253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459461703481699e-06,
+      "loss": 0.6243,
+      "step": 4254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455048447585286e-06,
+      "loss": 0.6672,
+      "step": 4255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45063675059794e-06,
+      "loss": 0.6089,
+      "step": 4256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446226613759958e-06,
+      "loss": 0.5891,
+      "step": 4257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441818038311215e-06,
+      "loss": 0.6313,
+      "step": 4258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437411025491131e-06,
+      "loss": 0.6887,
+      "step": 4259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433005576538699e-06,
+      "loss": 0.6672,
+      "step": 4260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428601692692469e-06,
+      "loss": 0.6196,
+      "step": 4261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424199375190552e-06,
+      "loss": 0.6166,
+      "step": 4262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4197986252706135e-06,
+      "loss": 0.5764,
+      "step": 4263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415399444169878e-06,
+      "loss": 0.5991,
+      "step": 4264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411001833125139e-06,
+      "loss": 0.5337,
+      "step": 4265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406605793372736e-06,
+      "loss": 0.6321,
+      "step": 4266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402211326148579e-06,
+      "loss": 0.6135,
+      "step": 4267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397818432688122e-06,
+      "loss": 0.562,
+      "step": 4268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.393427114226393e-06,
+      "loss": 0.6047,
+      "step": 4269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389037371997958e-06,
+      "loss": 0.5066,
+      "step": 4270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384649207236958e-06,
+      "loss": 0.636,
+      "step": 4271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380262621177081e-06,
+      "loss": 0.6736,
+      "step": 4272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375877615051575e-06,
+      "loss": 0.625,
+      "step": 4273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371494190093235e-06,
+      "loss": 0.5813,
+      "step": 4274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367112347534427e-06,
+      "loss": 0.5339,
+      "step": 4275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362732088607059e-06,
+      "loss": 0.7124,
+      "step": 4276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358353414542594e-06,
+      "loss": 0.5339,
+      "step": 4277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3539763265720615e-06,
+      "loss": 0.6509,
+      "step": 4278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349600825926028e-06,
+      "loss": 0.7083,
+      "step": 4279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345226913834632e-06,
+      "loss": 0.7102,
+      "step": 4280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3408545915275494e-06,
+      "loss": 0.5938,
+      "step": 4281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336483860234022e-06,
+      "loss": 0.5555,
+      "step": 4282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332114721182831e-06,
+      "loss": 0.502,
+      "step": 4283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327747175602321e-06,
+      "loss": 0.6505,
+      "step": 4284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323381224720384e-06,
+      "loss": 0.751,
+      "step": 4285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319016869764471e-06,
+      "loss": 0.5669,
+      "step": 4286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314654111961573e-06,
+      "loss": 0.6582,
+      "step": 4287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310292952538231e-06,
+      "loss": 0.7456,
+      "step": 4288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305933392720551e-06,
+      "loss": 0.5734,
+      "step": 4289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301575433734174e-06,
+      "loss": 0.7256,
+      "step": 4290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2972190768043055e-06,
+      "loss": 0.6536,
+      "step": 4291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.6868,
+      "step": 4292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2885111740126155e-06,
+      "loss": 0.6816,
+      "step": 4293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284159630598935e-06,
+      "loss": 0.6418,
+      "step": 4294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279809694138042e-06,
+      "loss": 0.6028,
+      "step": 4295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275461365852879e-06,
+      "loss": 0.6938,
+      "step": 4296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271114646965939e-06,
+      "loss": 0.6738,
+      "step": 4297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266769538699253e-06,
+      "loss": 0.7,
+      "step": 4298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262426042274413e-06,
+      "loss": 0.5996,
+      "step": 4299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258084158912545e-06,
+      "loss": 0.7505,
+      "step": 4300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253743889834327e-06,
+      "loss": 0.6587,
+      "step": 4301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249405236259988e-06,
+      "loss": 0.699,
+      "step": 4302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245068199409292e-06,
+      "loss": 0.665,
+      "step": 4303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240732780501563e-06,
+      "loss": 0.7073,
+      "step": 4304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2363989807556515e-06,
+      "loss": 0.6924,
+      "step": 4305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232066801389969e-06,
+      "loss": 0.6243,
+      "step": 4306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227736243622468e-06,
+      "loss": 0.6538,
+      "step": 4307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2234073086706366e-06,
+      "loss": 0.7217,
+      "step": 4308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219079997751515e-06,
+      "loss": 0.4678,
+      "step": 4309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214754312081688e-06,
+      "loss": 0.5476,
+      "step": 4310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2104302528772775e-06,
+      "loss": 0.6077,
+      "step": 4311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206107821353946e-06,
+      "loss": 0.5803,
+      "step": 4312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20178701872691e-06,
+      "loss": 0.6827,
+      "step": 4313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197467846210917e-06,
+      "loss": 0.5559,
+      "step": 4314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1931503050202614e-06,
+      "loss": 0.666,
+      "step": 4315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188834396368776e-06,
+      "loss": 0.552,
+      "step": 4316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184520121469841e-06,
+      "loss": 0.5515,
+      "step": 4317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1802074815363656e-06,
+      "loss": 0.7524,
+      "step": 4318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175896477780812e-06,
+      "loss": 0.6501,
+      "step": 4319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171587111415175e-06,
+      "loss": 0.5939,
+      "step": 4320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167279383650995e-06,
+      "loss": 0.7073,
+      "step": 4321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162973295699339e-06,
+      "loss": 0.7168,
+      "step": 4322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158668848770832e-06,
+      "loss": 0.5901,
+      "step": 4323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154366044075623e-06,
+      "loss": 0.5854,
+      "step": 4324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150064882823397e-06,
+      "loss": 0.6267,
+      "step": 4325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145765366223393e-06,
+      "loss": 0.6543,
+      "step": 4326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141467495484371e-06,
+      "loss": 0.6316,
+      "step": 4327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137171271814644e-06,
+      "loss": 0.6172,
+      "step": 4328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132876696422043e-06,
+      "loss": 0.6042,
+      "step": 4329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128583770513951e-06,
+      "loss": 0.415,
+      "step": 4330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124292495297287e-06,
+      "loss": 0.7178,
+      "step": 4331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120002871978491e-06,
+      "loss": 0.5581,
+      "step": 4332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115714901763557e-06,
+      "loss": 0.6646,
+      "step": 4333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111428585858005e-06,
+      "loss": 0.6313,
+      "step": 4334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107143925466886e-06,
+      "loss": 0.7422,
+      "step": 4335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102860921794798e-06,
+      "loss": 0.6929,
+      "step": 4336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098579576045861e-06,
+      "loss": 0.7625,
+      "step": 4337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0942998894237315e-06,
+      "loss": 0.6829,
+      "step": 4338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090021863131608e-06,
+      "loss": 0.7239,
+      "step": 4339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085745498372208e-06,
+      "loss": 0.7236,
+      "step": 4340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081470796347797e-06,
+      "loss": 0.665,
+      "step": 4341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07719775826016e-06,
+      "loss": 0.6511,
+      "step": 4342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072926385310621e-06,
+      "loss": 0.5693,
+      "step": 4343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068656678700039e-06,
+      "loss": 0.783,
+      "step": 4344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0643886396288e-06,
+      "loss": 0.7385,
+      "step": 4345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060122269296817e-06,
+      "loss": 0.7339,
+      "step": 4346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055857568903544e-06,
+      "loss": 0.7705,
+      "step": 4347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051594539647956e-06,
+      "loss": 0.686,
+      "step": 4348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0473331827285615e-06,
+      "loss": 0.7246,
+      "step": 4349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043073499343403e-06,
+      "loss": 0.575,
+      "step": 4350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038815490690043e-06,
+      "loss": 0.7119,
+      "step": 4351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034559157965587e-06,
+      "loss": 0.6606,
+      "step": 4352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030304502366654e-06,
+      "loss": 0.6562,
+      "step": 4353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026051525089403e-06,
+      "loss": 0.6553,
+      "step": 4354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021800227329516e-06,
+      "loss": 0.5413,
+      "step": 4355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017550610282208e-06,
+      "loss": 0.7356,
+      "step": 4356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01330267514221e-06,
+      "loss": 0.6792,
+      "step": 4357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009056423103796e-06,
+      "loss": 0.7004,
+      "step": 4358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0048118553607485e-06,
+      "loss": 0.4723,
+      "step": 4359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000568973106396e-06,
+      "loss": 0.6404,
+      "step": 4360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996327777533579e-06,
+      "loss": 0.6113,
+      "step": 4361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9920882698346645e-06,
+      "loss": 0.7759,
+      "step": 4362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987850451201556e-06,
+      "loss": 0.6951,
+      "step": 4363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983614322825668e-06,
+      "loss": 0.5247,
+      "step": 4364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9793798858979525e-06,
+      "loss": 0.634,
+      "step": 4365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975147141608876e-06,
+      "loss": 0.637,
+      "step": 4366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9709160911484345e-06,
+      "loss": 0.5447,
+      "step": 4367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966686735706146e-06,
+      "loss": 0.5693,
+      "step": 4368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962459076471059e-06,
+      "loss": 0.7827,
+      "step": 4369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958233114631729e-06,
+      "loss": 0.6316,
+      "step": 4370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 0.7854,
+      "step": 4371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949786287892235e-06,
+      "loss": 0.7295,
+      "step": 4372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945565425366806e-06,
+      "loss": 0.5459,
+      "step": 4373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941346264986627e-06,
+      "loss": 0.5859,
+      "step": 4374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937128807937869e-06,
+      "loss": 0.6926,
+      "step": 4375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9329130554062324e-06,
+      "loss": 0.6343,
+      "step": 4376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9286990085769285e-06,
+      "loss": 0.6475,
+      "step": 4377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9244866686347e-06,
+      "loss": 0.5852,
+      "step": 4378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920276036763807e-06,
+      "loss": 0.6832,
+      "step": 4379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916067114148028e-06,
+      "loss": 0.5415,
+      "step": 4380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911859901970654e-06,
+      "loss": 0.5618,
+      "step": 4381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90765440141451e-06,
+      "loss": 0.7285,
+      "step": 4382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903450613661924e-06,
+      "loss": 0.5461,
+      "step": 4383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899248539894756e-06,
+      "loss": 0.6078,
+      "step": 4384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895048181294377e-06,
+      "loss": 0.533,
+      "step": 4385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890849539041671e-06,
+      "loss": 0.666,
+      "step": 4386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886652614317053e-06,
+      "loss": 0.5544,
+      "step": 4387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882457408300441e-06,
+      "loss": 0.5549,
+      "step": 4388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878263922171281e-06,
+      "loss": 0.5613,
+      "step": 4389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874072157108527e-06,
+      "loss": 0.6152,
+      "step": 4390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869882114290655e-06,
+      "loss": 0.5624,
+      "step": 4391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865693794895654e-06,
+      "loss": 0.6543,
+      "step": 4392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861507200101033e-06,
+      "loss": 0.7778,
+      "step": 4393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857322331083807e-06,
+      "loss": 0.678,
+      "step": 4394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853139189020515e-06,
+      "loss": 0.6792,
+      "step": 4395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848957775087207e-06,
+      "loss": 0.5605,
+      "step": 4396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844778090459439e-06,
+      "loss": 0.4929,
+      "step": 4397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840600136312298e-06,
+      "loss": 0.6768,
+      "step": 4398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8364239138203676e-06,
+      "loss": 0.6348,
+      "step": 4399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832249424157759e-06,
+      "loss": 0.6047,
+      "step": 4400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828076668498083e-06,
+      "loss": 0.5676,
+      "step": 4401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823905648014472e-06,
+      "loss": 0.6587,
+      "step": 4402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819736363879567e-06,
+      "loss": 0.587,
+      "step": 4403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815568817265527e-06,
+      "loss": 0.5674,
+      "step": 4404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811403009344008e-06,
+      "loss": 0.5842,
+      "step": 4405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8072389412861953e-06,
+      "loss": 0.6263,
+      "step": 4406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803076614262767e-06,
+      "loss": 0.5598,
+      "step": 4407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7989160294439297e-06,
+      "loss": 0.676,
+      "step": 4408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794757187999386e-06,
+      "loss": 0.5814,
+      "step": 4409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.790600091098351e-06,
+      "loss": 0.6167,
+      "step": 4410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7864447399095584e-06,
+      "loss": 0.6499,
+      "step": 4411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782291135601239e-06,
+      "loss": 0.6055,
+      "step": 4412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778139279341141e-06,
+      "loss": 0.6177,
+      "step": 4413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7739891722965207e-06,
+      "loss": 0.582,
+      "step": 4414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769840815634135e-06,
+      "loss": 0.6943,
+      "step": 4415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7656942105202554e-06,
+      "loss": 0.7937,
+      "step": 4416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7615493581206654e-06,
+      "loss": 0.6134,
+      "step": 4417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.757406259600641e-06,
+      "loss": 0.7397,
+      "step": 4418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7532649161249813e-06,
+      "loss": 0.6899,
+      "step": 4419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749125328857981e-06,
+      "loss": 0.6577,
+      "step": 4420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7449874989634427e-06,
+      "loss": 0.4786,
+      "step": 4421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7408514276046826e-06,
+      "loss": 0.7844,
+      "step": 4422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736717115944509e-06,
+      "loss": 0.7935,
+      "step": 4423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732584565145252e-06,
+      "loss": 0.5981,
+      "step": 4424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7284537763687312e-06,
+      "loss": 0.7,
+      "step": 4425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7243247507762793e-06,
+      "loss": 0.7534,
+      "step": 4426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720197489528732e-06,
+      "loss": 0.6227,
+      "step": 4427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716071993786433e-06,
+      "loss": 0.5464,
+      "step": 4428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7119482647092187e-06,
+      "loss": 0.5151,
+      "step": 4429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70782630345644e-06,
+      "loss": 0.512,
+      "step": 4430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703706111186941e-06,
+      "loss": 0.6353,
+      "step": 4431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69958768905908e-06,
+      "loss": 0.6707,
+      "step": 4432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695471038230708e-06,
+      "loss": 0.5573,
+      "step": 4433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6913561598591775e-06,
+      "loss": 0.52,
+      "step": 4434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6872430551013526e-06,
+      "loss": 0.5205,
+      "step": 4435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6831317251135877e-06,
+      "loss": 0.6763,
+      "step": 4436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6790221710517447e-06,
+      "loss": 0.6558,
+      "step": 4437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6749143940711885e-06,
+      "loss": 0.7117,
+      "step": 4438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670808395326775e-06,
+      "loss": 0.7861,
+      "step": 4439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6667041759728673e-06,
+      "loss": 0.9141,
+      "step": 4440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6626017371633314e-06,
+      "loss": 0.6614,
+      "step": 4441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65850108005152e-06,
+      "loss": 0.5559,
+      "step": 4442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6544022057903008e-06,
+      "loss": 0.7039,
+      "step": 4443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650305115532028e-06,
+      "loss": 0.5234,
+      "step": 4444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646209810428556e-06,
+      "loss": 0.483,
+      "step": 4445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642116291631246e-06,
+      "loss": 0.616,
+      "step": 4446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6380245602909447e-06,
+      "loss": 0.5095,
+      "step": 4447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633934617558008e-06,
+      "loss": 0.7314,
+      "step": 4448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6298464645822783e-06,
+      "loss": 0.6113,
+      "step": 4449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.7034,
+      "step": 4450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6216755324993223e-06,
+      "loss": 0.6611,
+      "step": 4451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6175927556892764e-06,
+      "loss": 0.5232,
+      "step": 4452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6135117732307932e-06,
+      "loss": 0.7466,
+      "step": 4453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6094325862712055e-06,
+      "loss": 0.6997,
+      "step": 4454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605355195957332e-06,
+      "loss": 0.5228,
+      "step": 4455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601279603435497e-06,
+      "loss": 0.6714,
+      "step": 4456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5972058098515094e-06,
+      "loss": 0.5557,
+      "step": 4457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5931338163506757e-06,
+      "loss": 0.6255,
+      "step": 4458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589063624077802e-06,
+      "loss": 0.6443,
+      "step": 4459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5849952341771767e-06,
+      "loss": 0.5586,
+      "step": 4460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580928647792592e-06,
+      "loss": 0.5884,
+      "step": 4461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.576863866067328e-06,
+      "loss": 0.6431,
+      "step": 4462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5728008901441637e-06,
+      "loss": 0.6787,
+      "step": 4463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568739721165355e-06,
+      "loss": 0.595,
+      "step": 4464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56468036027267e-06,
+      "loss": 0.521,
+      "step": 4465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56062280860735e-06,
+      "loss": 0.5818,
+      "step": 4466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5565670673101437e-06,
+      "loss": 0.5688,
+      "step": 4467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5525131375212797e-06,
+      "loss": 0.6682,
+      "step": 4468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548461020380477e-06,
+      "loss": 0.6599,
+      "step": 4469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.544410717026956e-06,
+      "loss": 0.6951,
+      "step": 4470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540362228599412e-06,
+      "loss": 0.5237,
+      "step": 4471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5363155562360474e-06,
+      "loss": 0.7585,
+      "step": 4472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532270701074535e-06,
+      "loss": 0.7327,
+      "step": 4473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528227664252052e-06,
+      "loss": 0.5962,
+      "step": 4474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524186446905258e-06,
+      "loss": 0.6538,
+      "step": 4475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5201470501703038e-06,
+      "loss": 0.7725,
+      "step": 4476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5161094751828217e-06,
+      "loss": 0.5474,
+      "step": 4477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5120737230779434e-06,
+      "loss": 0.5986,
+      "step": 4478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5080397949902733e-06,
+      "loss": 0.7297,
+      "step": 4479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504007692053918e-06,
+      "loss": 0.5322,
+      "step": 4480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4999774154024604e-06,
+      "loss": 0.7944,
+      "step": 4481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4959489661689693e-06,
+      "loss": 0.4172,
+      "step": 4482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491922345486013e-06,
+      "loss": 0.6353,
+      "step": 4483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4878975544856285e-06,
+      "loss": 0.5583,
+      "step": 4484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48387459429935e-06,
+      "loss": 0.5806,
+      "step": 4485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479853466058194e-06,
+      "loss": 0.8018,
+      "step": 4486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4758341708926634e-06,
+      "loss": 0.7314,
+      "step": 4487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471816709932738e-06,
+      "loss": 0.6123,
+      "step": 4488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467801084307896e-06,
+      "loss": 0.7017,
+      "step": 4489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463787295147083e-06,
+      "loss": 0.5515,
+      "step": 4490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459775343578743e-06,
+      "loss": 0.5852,
+      "step": 4491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.455765230730794e-06,
+      "loss": 0.5892,
+      "step": 4492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4517569577306385e-06,
+      "loss": 0.7566,
+      "step": 4493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447750525705168e-06,
+      "loss": 0.5969,
+      "step": 4494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.443745935780746e-06,
+      "loss": 0.6074,
+      "step": 4495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43974318908323e-06,
+      "loss": 0.7372,
+      "step": 4496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4357422867379477e-06,
+      "loss": 0.6377,
+      "step": 4497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.431743229869716e-06,
+      "loss": 0.5997,
+      "step": 4498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4277460196028312e-06,
+      "loss": 0.7007,
+      "step": 4499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4237506570610734e-06,
+      "loss": 0.656,
+      "step": 4500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4197571433676935e-06,
+      "loss": 0.6067,
+      "step": 4501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4157654796454342e-06,
+      "loss": 0.6677,
+      "step": 4502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4117756670165083e-06,
+      "loss": 0.6477,
+      "step": 4503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4077877066026177e-06,
+      "loss": 0.751,
+      "step": 4504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403801599524935e-06,
+      "loss": 0.6853,
+      "step": 4505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399817346904113e-06,
+      "loss": 0.573,
+      "step": 4506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3958349498602926e-06,
+      "loss": 0.7046,
+      "step": 4507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3918544095130777e-06,
+      "loss": 0.687,
+      "step": 4508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.387875726981563e-06,
+      "loss": 0.625,
+      "step": 4509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383898903384316e-06,
+      "loss": 0.7166,
+      "step": 4510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3799239398393844e-06,
+      "loss": 0.6074,
+      "step": 4511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3759508374642857e-06,
+      "loss": 0.6409,
+      "step": 4512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3719795973760226e-06,
+      "loss": 0.717,
+      "step": 4513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368010220691067e-06,
+      "loss": 0.6616,
+      "step": 4514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3640427085253745e-06,
+      "loss": 0.6453,
+      "step": 4515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360077061994368e-06,
+      "loss": 0.6855,
+      "step": 4516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356113282212956e-06,
+      "loss": 0.6589,
+      "step": 4517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352151370295513e-06,
+      "loss": 0.6597,
+      "step": 4518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3481913273558885e-06,
+      "loss": 0.6987,
+      "step": 4519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3442331545074137e-06,
+      "loss": 0.6504,
+      "step": 4520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3402768528628935e-06,
+      "loss": 0.5674,
+      "step": 4521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336322423534597e-06,
+      "loss": 0.5361,
+      "step": 4522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.332369867634276e-06,
+      "loss": 0.5715,
+      "step": 4523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3284191862731585e-06,
+      "loss": 0.7336,
+      "step": 4524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3244703805619304e-06,
+      "loss": 0.7266,
+      "step": 4525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3205234516107686e-06,
+      "loss": 0.6401,
+      "step": 4526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316578400529307e-06,
+      "loss": 0.5558,
+      "step": 4527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3126352284266638e-06,
+      "loss": 0.7036,
+      "step": 4528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.7314,
+      "step": 4529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3047545255916303e-06,
+      "loss": 0.6011,
+      "step": 4530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3008169970748273e-06,
+      "loss": 0.6392,
+      "step": 4531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2968813519679998e-06,
+      "loss": 0.571,
+      "step": 4532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292947591377622e-06,
+      "loss": 0.5781,
+      "step": 4533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2890157164096315e-06,
+      "loss": 0.5701,
+      "step": 4534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285085728169439e-06,
+      "loss": 0.7261,
+      "step": 4535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2811576277619186e-06,
+      "loss": 0.7021,
+      "step": 4536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2772314162914198e-06,
+      "loss": 0.6797,
+      "step": 4537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2733070948617563e-06,
+      "loss": 0.5994,
+      "step": 4538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269384664576216e-06,
+      "loss": 0.6125,
+      "step": 4539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2654641265375474e-06,
+      "loss": 0.7598,
+      "step": 4540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2615454818479775e-06,
+      "loss": 0.6433,
+      "step": 4541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2576287316091925e-06,
+      "loss": 0.6101,
+      "step": 4542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253713876922343e-06,
+      "loss": 0.6453,
+      "step": 4543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2498009188880585e-06,
+      "loss": 0.665,
+      "step": 4544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.245889858606427e-06,
+      "loss": 0.718,
+      "step": 4545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2419806971770085e-06,
+      "loss": 0.6125,
+      "step": 4546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23807343569882e-06,
+      "loss": 0.5634,
+      "step": 4547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2341680752703554e-06,
+      "loss": 0.748,
+      "step": 4548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2302646169895634e-06,
+      "loss": 0.5211,
+      "step": 4549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226363061953868e-06,
+      "loss": 0.7954,
+      "step": 4550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2224634112601484e-06,
+      "loss": 0.6702,
+      "step": 4551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2185656660047592e-06,
+      "loss": 0.7163,
+      "step": 4552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21466982728351e-06,
+      "loss": 0.7383,
+      "step": 4553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2107758961916756e-06,
+      "loss": 0.6794,
+      "step": 4554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2068838738240017e-06,
+      "loss": 0.6731,
+      "step": 4555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2029937612746875e-06,
+      "loss": 0.6846,
+      "step": 4556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1991055596374032e-06,
+      "loss": 0.6104,
+      "step": 4557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1952192700052776e-06,
+      "loss": 0.5717,
+      "step": 4558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1913348934709076e-06,
+      "loss": 0.6782,
+      "step": 4559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1874524311263412e-06,
+      "loss": 0.5671,
+      "step": 4560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1835718840631e-06,
+      "loss": 0.665,
+      "step": 4561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179693253372157e-06,
+      "loss": 0.5239,
+      "step": 4562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.175816540143958e-06,
+      "loss": 0.7256,
+      "step": 4563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1719417454683954e-06,
+      "loss": 0.7842,
+      "step": 4564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1680688704348372e-06,
+      "loss": 0.6362,
+      "step": 4565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1641979161321e-06,
+      "loss": 0.7021,
+      "step": 4566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1603288836484638e-06,
+      "loss": 0.7036,
+      "step": 4567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156461774071672e-06,
+      "loss": 0.6397,
+      "step": 4568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152596588488923e-06,
+      "loss": 0.5908,
+      "step": 4569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1487333279868814e-06,
+      "loss": 0.6152,
+      "step": 4570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.144871993651657e-06,
+      "loss": 0.6416,
+      "step": 4571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1410125865688337e-06,
+      "loss": 0.6836,
+      "step": 4572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13715510782344e-06,
+      "loss": 0.7612,
+      "step": 4573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1332995584999748e-06,
+      "loss": 0.5698,
+      "step": 4574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1294459396823807e-06,
+      "loss": 0.6775,
+      "step": 4575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125594252454073e-06,
+      "loss": 0.7234,
+      "step": 4576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1217444978979105e-06,
+      "loss": 0.8403,
+      "step": 4577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117896677096214e-06,
+      "loss": 0.7344,
+      "step": 4578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114050791130766e-06,
+      "loss": 0.5499,
+      "step": 4579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1102068410827924e-06,
+      "loss": 0.7788,
+      "step": 4580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106364828032986e-06,
+      "loss": 0.5394,
+      "step": 4581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.102524753061491e-06,
+      "loss": 0.592,
+      "step": 4582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0986866172479123e-06,
+      "loss": 0.6021,
+      "step": 4583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094850421671295e-06,
+      "loss": 0.6238,
+      "step": 4584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910161674101567e-06,
+      "loss": 0.6943,
+      "step": 4585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087183855542453e-06,
+      "loss": 0.4685,
+      "step": 4586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0833534871456083e-06,
+      "loss": 0.605,
+      "step": 4587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0795250632964855e-06,
+      "loss": 0.5405,
+      "step": 4588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075698585071417e-06,
+      "loss": 0.6329,
+      "step": 4589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0718740535461757e-06,
+      "loss": 0.6162,
+      "step": 4590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0680514697959873e-06,
+      "loss": 0.6204,
+      "step": 4591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0642308348955384e-06,
+      "loss": 0.637,
+      "step": 4592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060412149918962e-06,
+      "loss": 0.6428,
+      "step": 4593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056595415939849e-06,
+      "loss": 0.5579,
+      "step": 4594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0527806340312293e-06,
+      "loss": 0.6323,
+      "step": 4595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048967805265598e-06,
+      "loss": 0.6631,
+      "step": 4596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04515693071489e-06,
+      "loss": 0.6587,
+      "step": 4597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041348011450499e-06,
+      "loss": 0.5857,
+      "step": 4598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037541048543261e-06,
+      "loss": 0.7317,
+      "step": 4599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033736043063473e-06,
+      "loss": 0.4946,
+      "step": 4600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029932996080872e-06,
+      "loss": 0.676,
+      "step": 4601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0261319086646446e-06,
+      "loss": 0.6855,
+      "step": 4602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0223327818834313e-06,
+      "loss": 0.5696,
+      "step": 4603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0185356168053225e-06,
+      "loss": 0.6025,
+      "step": 4604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01474041449785e-06,
+      "loss": 0.5984,
+      "step": 4605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0109471760279994e-06,
+      "loss": 0.8271,
+      "step": 4606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0071559024622078e-06,
+      "loss": 0.6143,
+      "step": 4607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 0.7297,
+      "step": 4608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999579254305748e-06,
+      "loss": 0.7812,
+      "step": 4609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9957938818451817e-06,
+      "loss": 0.6641,
+      "step": 4610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9920104785488735e-06,
+      "loss": 0.6006,
+      "step": 4611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988229045480484e-06,
+      "loss": 0.6035,
+      "step": 4612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984449583703134e-06,
+      "loss": 0.644,
+      "step": 4613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9806720942793776e-06,
+      "loss": 0.4747,
+      "step": 4614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9768965782712177e-06,
+      "loss": 0.6055,
+      "step": 4615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731230367401044e-06,
+      "loss": 0.7871,
+      "step": 4616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.969351470746934e-06,
+      "loss": 0.7495,
+      "step": 4617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9655818813520475e-06,
+      "loss": 0.5806,
+      "step": 4618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961814269615223e-06,
+      "loss": 0.5195,
+      "step": 4619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.958048636595694e-06,
+      "loss": 0.6816,
+      "step": 4620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9542849833521235e-06,
+      "loss": 0.731,
+      "step": 4621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9505233109426346e-06,
+      "loss": 0.6924,
+      "step": 4622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9467636204247773e-06,
+      "loss": 0.7209,
+      "step": 4623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9430059128555586e-06,
+      "loss": 0.5371,
+      "step": 4624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9392501892914183e-06,
+      "loss": 0.7292,
+      "step": 4625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9354964507882376e-06,
+      "loss": 0.6992,
+      "step": 4626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9317446984013466e-06,
+      "loss": 0.6646,
+      "step": 4627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9279949331855175e-06,
+      "loss": 0.5557,
+      "step": 4628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9242471561949546e-06,
+      "loss": 0.647,
+      "step": 4629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9205013684833108e-06,
+      "loss": 0.5997,
+      "step": 4630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916757571103681e-06,
+      "loss": 0.6648,
+      "step": 4631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9130157651085924e-06,
+      "loss": 0.7063,
+      "step": 4632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9092759515500225e-06,
+      "loss": 0.6248,
+      "step": 4633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905538131479376e-06,
+      "loss": 0.7173,
+      "step": 4634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901802305947513e-06,
+      "loss": 0.7822,
+      "step": 4635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8980684760047182e-06,
+      "loss": 0.6765,
+      "step": 4636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894336642700726e-06,
+      "loss": 0.627,
+      "step": 4637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8906068070847026e-06,
+      "loss": 0.6052,
+      "step": 4638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8868789702052513e-06,
+      "loss": 0.5374,
+      "step": 4639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8831531331104212e-06,
+      "loss": 0.751,
+      "step": 4640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8794292968476956e-06,
+      "loss": 0.7285,
+      "step": 4641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8757074624639958e-06,
+      "loss": 0.5437,
+      "step": 4642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8719876310056738e-06,
+      "loss": 0.4912,
+      "step": 4643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8682698035185307e-06,
+      "loss": 0.5143,
+      "step": 4644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8645539810477916e-06,
+      "loss": 0.7104,
+      "step": 4645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.860840164638129e-06,
+      "loss": 0.6677,
+      "step": 4646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8571283553336405e-06,
+      "loss": 0.7676,
+      "step": 4647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8534185541778712e-06,
+      "loss": 0.4705,
+      "step": 4648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8497107622137923e-06,
+      "loss": 0.6555,
+      "step": 4649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8460049804838098e-06,
+      "loss": 0.6345,
+      "step": 4650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842301210029772e-06,
+      "loss": 0.5876,
+      "step": 4651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838599451892957e-06,
+      "loss": 0.5994,
+      "step": 4652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8348997071140827e-06,
+      "loss": 0.6021,
+      "step": 4653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8312019767332888e-06,
+      "loss": 0.7058,
+      "step": 4654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8275062617901617e-06,
+      "loss": 0.574,
+      "step": 4655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8238125633237113e-06,
+      "loss": 0.6489,
+      "step": 4656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820120882372389e-06,
+      "loss": 0.6438,
+      "step": 4657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816431219974072e-06,
+      "loss": 0.7368,
+      "step": 4658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812743577166075e-06,
+      "loss": 0.6658,
+      "step": 4659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8090579549851384e-06,
+      "loss": 0.7566,
+      "step": 4660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805374354467446e-06,
+      "loss": 0.7949,
+      "step": 4661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801692776648602e-06,
+      "loss": 0.6492,
+      "step": 4662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7980132225636437e-06,
+      "loss": 0.7241,
+      "step": 4663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7943356932470446e-06,
+      "loss": 0.5952,
+      "step": 4664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7906601897327057e-06,
+      "loss": 0.748,
+      "step": 4665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786986713053963e-06,
+      "loss": 0.6179,
+      "step": 4666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7833152642435724e-06,
+      "loss": 0.6335,
+      "step": 4667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7796458443337304e-06,
+      "loss": 0.6105,
+      "step": 4668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.775978454356055e-06,
+      "loss": 0.6169,
+      "step": 4669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7723130953416023e-06,
+      "loss": 0.6499,
+      "step": 4670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768649768320845e-06,
+      "loss": 0.7544,
+      "step": 4671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7649884743236997e-06,
+      "loss": 0.823,
+      "step": 4672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7613292143794958e-06,
+      "loss": 0.5234,
+      "step": 4673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757671989517006e-06,
+      "loss": 0.6562,
+      "step": 4674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540168007644152e-06,
+      "loss": 0.759,
+      "step": 4675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7503636491493502e-06,
+      "loss": 0.5292,
+      "step": 4676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74671253569886e-06,
+      "loss": 0.6746,
+      "step": 4677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7430634614394135e-06,
+      "loss": 0.6719,
+      "step": 4678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739416427396919e-06,
+      "loss": 0.6616,
+      "step": 4679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735771434596698e-06,
+      "loss": 0.6914,
+      "step": 4680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.732128484063512e-06,
+      "loss": 0.6925,
+      "step": 4681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7284875768215335e-06,
+      "loss": 0.5562,
+      "step": 4682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7248487138943738e-06,
+      "loss": 0.8464,
+      "step": 4683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721211896305059e-06,
+      "loss": 0.6311,
+      "step": 4684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.717577125076051e-06,
+      "loss": 0.533,
+      "step": 4685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713944401229226e-06,
+      "loss": 0.8062,
+      "step": 4686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.7544,
+      "step": 4687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706685099766765e-06,
+      "loss": 0.7086,
+      "step": 4688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703058524192014e-06,
+      "loss": 0.562,
+      "step": 4689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6994340000812123e-06,
+      "loss": 0.6484,
+      "step": 4690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695811528453355e-06,
+      "loss": 0.6404,
+      "step": 4691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6921911103268683e-06,
+      "loss": 0.7825,
+      "step": 4692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6885727467195943e-06,
+      "loss": 0.6304,
+      "step": 4693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6849564386488047e-06,
+      "loss": 0.6599,
+      "step": 4694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681342187131183e-06,
+      "loss": 0.7197,
+      "step": 4695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6777299931828482e-06,
+      "loss": 0.7729,
+      "step": 4696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674119857819326e-06,
+      "loss": 0.6675,
+      "step": 4697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6705117820555783e-06,
+      "loss": 0.5701,
+      "step": 4698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6669057669059727e-06,
+      "loss": 0.6736,
+      "step": 4699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6633018133843104e-06,
+      "loss": 0.6978,
+      "step": 4700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6596999225038076e-06,
+      "loss": 0.6897,
+      "step": 4701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6561000952770987e-06,
+      "loss": 0.5955,
+      "step": 4702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6525023327162426e-06,
+      "loss": 0.6384,
+      "step": 4703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648906635832711e-06,
+      "loss": 0.8081,
+      "step": 4704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6453130056374054e-06,
+      "loss": 0.6616,
+      "step": 4705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6417214431406325e-06,
+      "loss": 0.6516,
+      "step": 4706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6381319493521306e-06,
+      "loss": 0.783,
+      "step": 4707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6345445252810476e-06,
+      "loss": 0.7773,
+      "step": 4708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6309591719359563e-06,
+      "loss": 0.6284,
+      "step": 4709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627375890324838e-06,
+      "loss": 0.7358,
+      "step": 4710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623794681455105e-06,
+      "loss": 0.7659,
+      "step": 4711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620215546333572e-06,
+      "loss": 0.6777,
+      "step": 4712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6166384859664818e-06,
+      "loss": 0.7295,
+      "step": 4713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613063501359492e-06,
+      "loss": 0.7012,
+      "step": 4714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6094905935176682e-06,
+      "loss": 0.5986,
+      "step": 4715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605919763445507e-06,
+      "loss": 0.6611,
+      "step": 4716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6023510121469043e-06,
+      "loss": 0.6237,
+      "step": 4717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5987843406251865e-06,
+      "loss": 0.7795,
+      "step": 4718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595219749883083e-06,
+      "loss": 0.6243,
+      "step": 4719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5916572409227492e-06,
+      "loss": 0.6646,
+      "step": 4720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588096814745743e-06,
+      "loss": 0.6997,
+      "step": 4721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5845384723530507e-06,
+      "loss": 0.6992,
+      "step": 4722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580982214745059e-06,
+      "loss": 0.454,
+      "step": 4723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577428042921578e-06,
+      "loss": 0.7599,
+      "step": 4724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5738759578818316e-06,
+      "loss": 0.6531,
+      "step": 4725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570325960624448e-06,
+      "loss": 0.7388,
+      "step": 4726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5667780521474804e-06,
+      "loss": 0.6885,
+      "step": 4727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5632322334483826e-06,
+      "loss": 0.7036,
+      "step": 4728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559688505524033e-06,
+      "loss": 0.7329,
+      "step": 4729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556146869370709e-06,
+      "loss": 0.6699,
+      "step": 4730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5526073259841156e-06,
+      "loss": 0.5432,
+      "step": 4731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5490698763593523e-06,
+      "loss": 0.7129,
+      "step": 4732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5455345214909456e-06,
+      "loss": 0.6096,
+      "step": 4733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.542001262372821e-06,
+      "loss": 0.6648,
+      "step": 4734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538470099998325e-06,
+      "loss": 0.7048,
+      "step": 4735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534941035360202e-06,
+      "loss": 0.7236,
+      "step": 4736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5314140694506195e-06,
+      "loss": 0.634,
+      "step": 4737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527889203261151e-06,
+      "loss": 0.7683,
+      "step": 4738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5243664377827737e-06,
+      "loss": 0.6052,
+      "step": 4739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5208457740058834e-06,
+      "loss": 0.7744,
+      "step": 4740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517327212920275e-06,
+      "loss": 0.6167,
+      "step": 4741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5138107555151634e-06,
+      "loss": 0.6023,
+      "step": 4742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510296402779161e-06,
+      "loss": 0.6396,
+      "step": 4743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5067841557002994e-06,
+      "loss": 0.6177,
+      "step": 4744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503274015266006e-06,
+      "loss": 0.7898,
+      "step": 4745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4997659824631295e-06,
+      "loss": 0.7483,
+      "step": 4746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496260058277912e-06,
+      "loss": 0.6372,
+      "step": 4747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492756243696014e-06,
+      "loss": 0.6267,
+      "step": 4748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4892545397025016e-06,
+      "loss": 0.678,
+      "step": 4749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4857549472818376e-06,
+      "loss": 0.6501,
+      "step": 4750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4822574674179056e-06,
+      "loss": 0.6213,
+      "step": 4751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478762101093982e-06,
+      "loss": 0.5498,
+      "step": 4752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4752688492927602e-06,
+      "loss": 0.7505,
+      "step": 4753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4717777129963283e-06,
+      "loss": 0.605,
+      "step": 4754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4682886931861915e-06,
+      "loss": 0.5023,
+      "step": 4755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4648017908432463e-06,
+      "loss": 0.7051,
+      "step": 4756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4613170069478097e-06,
+      "loss": 0.6899,
+      "step": 4757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.457834342479587e-06,
+      "loss": 0.6082,
+      "step": 4758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.454353798417698e-06,
+      "loss": 0.6536,
+      "step": 4759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450875375740668e-06,
+      "loss": 0.7102,
+      "step": 4760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4473990754264144e-06,
+      "loss": 0.6641,
+      "step": 4761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4439248984522723e-06,
+      "loss": 0.7754,
+      "step": 4762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440452845794966e-06,
+      "loss": 0.6772,
+      "step": 4763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436982918430636e-06,
+      "loss": 0.7341,
+      "step": 4764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4335151173348115e-06,
+      "loss": 0.7227,
+      "step": 4765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 0.7351,
+      "step": 4766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4265858978478484e-06,
+      "loss": 0.478,
+      "step": 4767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4231244814047914e-06,
+      "loss": 0.7827,
+      "step": 4768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419665195126406e-06,
+      "loss": 0.6819,
+      "step": 4769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4162080399852407e-06,
+      "loss": 0.7842,
+      "step": 4770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4127530169532363e-06,
+      "loss": 0.696,
+      "step": 4771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4093001270017424e-06,
+      "loss": 0.8208,
+      "step": 4772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4058493711015063e-06,
+      "loss": 0.6211,
+      "step": 4773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402400750222669e-06,
+      "loss": 0.5798,
+      "step": 4774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3989542653347832e-06,
+      "loss": 0.5876,
+      "step": 4775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3955099174067885e-06,
+      "loss": 0.6484,
+      "step": 4776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3920677074070364e-06,
+      "loss": 0.7588,
+      "step": 4777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3886276363032634e-06,
+      "loss": 0.5281,
+      "step": 4778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3851897050626185e-06,
+      "loss": 0.6616,
+      "step": 4779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3817539146516365e-06,
+      "loss": 0.6658,
+      "step": 4780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3783202660362613e-06,
+      "loss": 0.7363,
+      "step": 4781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374888760181826e-06,
+      "loss": 0.6426,
+      "step": 4782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3714593980530666e-06,
+      "loss": 0.605,
+      "step": 4783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3680321806141182e-06,
+      "loss": 0.6361,
+      "step": 4784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364607108828504e-06,
+      "loss": 0.5986,
+      "step": 4785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3611841836591533e-06,
+      "loss": 0.7354,
+      "step": 4786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3577634060683853e-06,
+      "loss": 0.6096,
+      "step": 4787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3543447770179217e-06,
+      "loss": 0.6956,
+      "step": 4788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350928297468872e-06,
+      "loss": 0.7886,
+      "step": 4789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3475139683817516e-06,
+      "loss": 0.4951,
+      "step": 4790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3441017907164608e-06,
+      "loss": 0.6245,
+      "step": 4791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340691765432305e-06,
+      "loss": 0.5103,
+      "step": 4792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337283893487974e-06,
+      "loss": 0.6086,
+      "step": 4793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333878175841563e-06,
+      "loss": 0.679,
+      "step": 4794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3304746134505517e-06,
+      "loss": 0.6372,
+      "step": 4795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3270732072718205e-06,
+      "loss": 0.5886,
+      "step": 4796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323673958261645e-06,
+      "loss": 0.6079,
+      "step": 4797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320276867375686e-06,
+      "loss": 0.7871,
+      "step": 4798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3168819355690054e-06,
+      "loss": 0.7136,
+      "step": 4799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3134891637960523e-06,
+      "loss": 0.6804,
+      "step": 4800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310098553010676e-06,
+      "loss": 0.6685,
+      "step": 4801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3067101041661066e-06,
+      "loss": 0.7051,
+      "step": 4802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303323818214982e-06,
+      "loss": 0.6917,
+      "step": 4803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2999396961093157e-06,
+      "loss": 0.6611,
+      "step": 4804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2965577388005266e-06,
+      "loss": 0.6396,
+      "step": 4805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2931779472394144e-06,
+      "loss": 0.8345,
+      "step": 4806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2898003223761756e-06,
+      "loss": 0.6167,
+      "step": 4807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2864248651604015e-06,
+      "loss": 0.6318,
+      "step": 4808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283051576541062e-06,
+      "loss": 0.7134,
+      "step": 4809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2796804574665298e-06,
+      "loss": 0.5603,
+      "step": 4810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2763115088845565e-06,
+      "loss": 0.717,
+      "step": 4811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272944731742296e-06,
+      "loss": 0.7583,
+      "step": 4812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269580126986277e-06,
+      "loss": 0.5781,
+      "step": 4813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2662176955624336e-06,
+      "loss": 0.7285,
+      "step": 4814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262857438416074e-06,
+      "loss": 0.6653,
+      "step": 4815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2594993564919067e-06,
+      "loss": 0.6519,
+      "step": 4816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2561434507340173e-06,
+      "loss": 0.8176,
+      "step": 4817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2527897220858943e-06,
+      "loss": 0.4977,
+      "step": 4818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249438171490398e-06,
+      "loss": 0.6677,
+      "step": 4819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2460887998897875e-06,
+      "loss": 0.6812,
+      "step": 4820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242741608225708e-06,
+      "loss": 0.6821,
+      "step": 4821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239396597439186e-06,
+      "loss": 0.7058,
+      "step": 4822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236053768470644e-06,
+      "loss": 0.5603,
+      "step": 4823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2327131222598784e-06,
+      "loss": 0.5436,
+      "step": 4824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2293746597460864e-06,
+      "loss": 0.6768,
+      "step": 4825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2260383818678388e-06,
+      "loss": 0.7039,
+      "step": 4826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2227042895631013e-06,
+      "loss": 0.4871,
+      "step": 4827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2193723837692183e-06,
+      "loss": 0.686,
+      "step": 4828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2160426654229262e-06,
+      "loss": 0.5247,
+      "step": 4829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2127151354603395e-06,
+      "loss": 0.7903,
+      "step": 4830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2093897948169607e-06,
+      "loss": 0.697,
+      "step": 4831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2060666444276814e-06,
+      "loss": 0.7214,
+      "step": 4832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202745685226768e-06,
+      "loss": 0.8086,
+      "step": 4833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19942691814788e-06,
+      "loss": 0.6572,
+      "step": 4834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1961103441240516e-06,
+      "loss": 0.7041,
+      "step": 4835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1927959640877096e-06,
+      "loss": 0.5906,
+      "step": 4836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189483778970656e-06,
+      "loss": 0.6006,
+      "step": 4837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1861737897040835e-06,
+      "loss": 0.5454,
+      "step": 4838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1828659972185583e-06,
+      "loss": 0.5662,
+      "step": 4839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1795604024440397e-06,
+      "loss": 0.7847,
+      "step": 4840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1762570063098577e-06,
+      "loss": 0.7847,
+      "step": 4841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1729558097447333e-06,
+      "loss": 0.8501,
+      "step": 4842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169656813676767e-06,
+      "loss": 0.7356,
+      "step": 4843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1663600190334346e-06,
+      "loss": 0.6179,
+      "step": 4844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 0.6589,
+      "step": 4845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1597730377275106e-06,
+      "loss": 0.6125,
+      "step": 4846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1564828529167836e-06,
+      "loss": 0.7786,
+      "step": 4847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1531948732344222e-06,
+      "loss": 0.5657,
+      "step": 4848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1499090996048146e-06,
+      "loss": 0.614,
+      "step": 4849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146625532951717e-06,
+      "loss": 0.5952,
+      "step": 4850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1433441741982796e-06,
+      "loss": 0.6973,
+      "step": 4851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1400650242670185e-06,
+      "loss": 0.6284,
+      "step": 4852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1367880840798405e-06,
+      "loss": 0.5999,
+      "step": 4853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1335133545580176e-06,
+      "loss": 0.7307,
+      "step": 4854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1302408366222137e-06,
+      "loss": 0.6423,
+      "step": 4855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126970531192467e-06,
+      "loss": 0.5994,
+      "step": 4856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1237024391881855e-06,
+      "loss": 0.6079,
+      "step": 4857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1204365615281687e-06,
+      "loss": 0.6729,
+      "step": 4858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1171728991305797e-06,
+      "loss": 0.8186,
+      "step": 4859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1139114529129713e-06,
+      "loss": 0.6604,
+      "step": 4860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1106522237922612e-06,
+      "loss": 0.6313,
+      "step": 4861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1073952126847553e-06,
+      "loss": 0.6189,
+      "step": 4862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1041404205061255e-06,
+      "loss": 0.5598,
+      "step": 4863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100887848171431e-06,
+      "loss": 0.8418,
+      "step": 4864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097637496595094e-06,
+      "loss": 0.7046,
+      "step": 4865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0943893666909222e-06,
+      "loss": 0.6763,
+      "step": 4866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.091143459372099e-06,
+      "loss": 0.5745,
+      "step": 4867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087899775551174e-06,
+      "loss": 0.6436,
+      "step": 4868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0846583161400837e-06,
+      "loss": 0.5908,
+      "step": 4869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0814190820501245e-06,
+      "loss": 0.8003,
+      "step": 4870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0781820741919832e-06,
+      "loss": 0.7346,
+      "step": 4871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0749472934757066e-06,
+      "loss": 0.6211,
+      "step": 4872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0717147408107284e-06,
+      "loss": 0.6338,
+      "step": 4873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0684844171058426e-06,
+      "loss": 0.6057,
+      "step": 4874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065256323269229e-06,
+      "loss": 0.5889,
+      "step": 4875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0620304602084297e-06,
+      "loss": 0.6484,
+      "step": 4876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0588068288303696e-06,
+      "loss": 0.7446,
+      "step": 4877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0555854300413358e-06,
+      "loss": 0.6245,
+      "step": 4878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0523662647469954e-06,
+      "loss": 0.6396,
+      "step": 4879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491493338523893e-06,
+      "loss": 0.8491,
+      "step": 4880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0459346382619206e-06,
+      "loss": 0.7092,
+      "step": 4881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0427221788793736e-06,
+      "loss": 0.761,
+      "step": 4882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039511956607896e-06,
+      "loss": 0.6201,
+      "step": 4883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0363039723500155e-06,
+      "loss": 0.7446,
+      "step": 4884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0330982270076204e-06,
+      "loss": 0.6331,
+      "step": 4885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0298947214819796e-06,
+      "loss": 0.7075,
+      "step": 4886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026693456673722e-06,
+      "loss": 0.6973,
+      "step": 4887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0234944334828577e-06,
+      "loss": 0.5825,
+      "step": 4888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020297652808756e-06,
+      "loss": 0.5299,
+      "step": 4889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017103115550162e-06,
+      "loss": 0.6304,
+      "step": 4890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0139108226051927e-06,
+      "loss": 0.7585,
+      "step": 4891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0107207748713244e-06,
+      "loss": 0.5938,
+      "step": 4892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.007532973245413e-06,
+      "loss": 0.5676,
+      "step": 4893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0043474186236723e-06,
+      "loss": 0.7561,
+      "step": 4894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0011641119016957e-06,
+      "loss": 0.5228,
+      "step": 4895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997983053974434e-06,
+      "loss": 0.7458,
+      "step": 4896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9948042457362148e-06,
+      "loss": 0.8438,
+      "step": 4897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991627688080724e-06,
+      "loss": 0.5503,
+      "step": 4898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9884533819010252e-06,
+      "loss": 0.6284,
+      "step": 4899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985281328089539e-06,
+      "loss": 0.6338,
+      "step": 4900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982111527538062e-06,
+      "loss": 0.7603,
+      "step": 4901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9789439811377475e-06,
+      "loss": 0.6926,
+      "step": 4902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975778689779124e-06,
+      "loss": 0.7339,
+      "step": 4903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9726156543520834e-06,
+      "loss": 0.854,
+      "step": 4904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9694548757458776e-06,
+      "loss": 0.7173,
+      "step": 4905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966296354849133e-06,
+      "loss": 0.7021,
+      "step": 4906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9631400925498333e-06,
+      "loss": 0.708,
+      "step": 4907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959986089735336e-06,
+      "loss": 0.7083,
+      "step": 4908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9568343472923524e-06,
+      "loss": 0.7561,
+      "step": 4909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9536848661069697e-06,
+      "loss": 0.7139,
+      "step": 4910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9505376470646274e-06,
+      "loss": 0.6914,
+      "step": 4911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947392691050142e-06,
+      "loss": 0.6445,
+      "step": 4912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442499989476815e-06,
+      "loss": 0.6538,
+      "step": 4913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9411095716407855e-06,
+      "loss": 0.7261,
+      "step": 4914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937971410012358e-06,
+      "loss": 0.5674,
+      "step": 4915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9348355149446573e-06,
+      "loss": 0.6733,
+      "step": 4916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931701887319314e-06,
+      "loss": 0.7251,
+      "step": 4917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9285705280173115e-06,
+      "loss": 0.7329,
+      "step": 4918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925441437919008e-06,
+      "loss": 0.6882,
+      "step": 4919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9223146179041096e-06,
+      "loss": 0.6692,
+      "step": 4920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9191900688516973e-06,
+      "loss": 0.6631,
+      "step": 4921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9160677916402027e-06,
+      "loss": 0.7852,
+      "step": 4922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912947787147428e-06,
+      "loss": 0.7231,
+      "step": 4923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.7212,
+      "step": 4924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906714599826024e-06,
+      "loss": 0.6211,
+      "step": 4925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9036014187497965e-06,
+      "loss": 0.6998,
+      "step": 4926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9004905138970842e-06,
+      "loss": 0.7446,
+      "step": 4927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8973818861424931e-06,
+      "loss": 0.7349,
+      "step": 4928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8942755363599763e-06,
+      "loss": 0.7676,
+      "step": 4929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8911714654228608e-06,
+      "loss": 0.696,
+      "step": 4930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8880696742038184e-06,
+      "loss": 0.5894,
+      "step": 4931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884970163574895e-06,
+      "loss": 0.6594,
+      "step": 4932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8818729344074815e-06,
+      "loss": 0.7212,
+      "step": 4933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8787779875723389e-06,
+      "loss": 0.6301,
+      "step": 4934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8756853239395756e-06,
+      "loss": 0.741,
+      "step": 4935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8725949443786707e-06,
+      "loss": 0.504,
+      "step": 4936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8695068497584467e-06,
+      "loss": 0.646,
+      "step": 4937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8664210409470963e-06,
+      "loss": 0.7124,
+      "step": 4938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.863337518812165e-06,
+      "loss": 0.7085,
+      "step": 4939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8602562842205517e-06,
+      "loss": 0.6785,
+      "step": 4940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8571773380385194e-06,
+      "loss": 0.6128,
+      "step": 4941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8541006811316787e-06,
+      "loss": 0.6177,
+      "step": 4942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8510263143650076e-06,
+      "loss": 0.6204,
+      "step": 4943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8479542386028293e-06,
+      "loss": 0.5701,
+      "step": 4944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8448844547088307e-06,
+      "loss": 0.6855,
+      "step": 4945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8418169635460503e-06,
+      "loss": 0.6626,
+      "step": 4946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838751765976885e-06,
+      "loss": 0.5208,
+      "step": 4947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.835688862863082e-06,
+      "loss": 0.6958,
+      "step": 4948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8326282550657492e-06,
+      "loss": 0.5485,
+      "step": 4949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8295699434453485e-06,
+      "loss": 0.5056,
+      "step": 4950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8265139288616895e-06,
+      "loss": 0.5898,
+      "step": 4951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8234602121739453e-06,
+      "loss": 0.6729,
+      "step": 4952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8204087942406345e-06,
+      "loss": 0.7051,
+      "step": 4953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8173596759196388e-06,
+      "loss": 0.5938,
+      "step": 4954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8143128580681823e-06,
+      "loss": 0.6812,
+      "step": 4955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8112683415428534e-06,
+      "loss": 0.5391,
+      "step": 4956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8082261271995837e-06,
+      "loss": 0.8066,
+      "step": 4957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805186215893666e-06,
+      "loss": 0.6567,
+      "step": 4958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8021486084797368e-06,
+      "loss": 0.8354,
+      "step": 4959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7991133058117959e-06,
+      "loss": 0.7095,
+      "step": 4960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7960803087431833e-06,
+      "loss": 0.6497,
+      "step": 4961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930496181266e-06,
+      "loss": 0.679,
+      "step": 4962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7900212348140967e-06,
+      "loss": 0.6533,
+      "step": 4963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7869951596570677e-06,
+      "loss": 0.708,
+      "step": 4964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7839713935062719e-06,
+      "loss": 0.5654,
+      "step": 4965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809499372118056e-06,
+      "loss": 0.6216,
+      "step": 4966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7779307916231259e-06,
+      "loss": 0.8291,
+      "step": 4967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774913957589033e-06,
+      "loss": 0.6045,
+      "step": 4968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7718994359576835e-06,
+      "loss": 0.6792,
+      "step": 4969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768887227576578e-06,
+      "loss": 0.6555,
+      "step": 4970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7658773332925717e-06,
+      "loss": 0.6411,
+      "step": 4971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7628697539518647e-06,
+      "loss": 0.5322,
+      "step": 4972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7598644904000084e-06,
+      "loss": 0.6028,
+      "step": 4973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7568615434819059e-06,
+      "loss": 0.6189,
+      "step": 4974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7538609140418072e-06,
+      "loss": 0.8677,
+      "step": 4975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7508626029233088e-06,
+      "loss": 0.6975,
+      "step": 4976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7478666109693532e-06,
+      "loss": 0.7925,
+      "step": 4977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7448729390222408e-06,
+      "loss": 0.5312,
+      "step": 4978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7418815879236052e-06,
+      "loss": 0.6584,
+      "step": 4979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7388925585144434e-06,
+      "loss": 0.6572,
+      "step": 4980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7359058516350868e-06,
+      "loss": 0.7412,
+      "step": 4981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7329214681252214e-06,
+      "loss": 0.6438,
+      "step": 4982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7299394088238752e-06,
+      "loss": 0.709,
+      "step": 4983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7269596745694295e-06,
+      "loss": 0.7488,
+      "step": 4984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723982266199602e-06,
+      "loss": 0.5786,
+      "step": 4985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7210071845514641e-06,
+      "loss": 0.8123,
+      "step": 4986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7180344304614348e-06,
+      "loss": 0.6912,
+      "step": 4987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7150640047652688e-06,
+      "loss": 0.8955,
+      "step": 4988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120959082980782e-06,
+      "loss": 0.5891,
+      "step": 4989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7091301418943084e-06,
+      "loss": 0.5906,
+      "step": 4990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061667063877607e-06,
+      "loss": 0.6572,
+      "step": 4991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032056026115729e-06,
+      "loss": 0.5505,
+      "step": 4992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7002468313982334e-06,
+      "loss": 0.7688,
+      "step": 4993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6972903935795681e-06,
+      "loss": 0.7222,
+      "step": 4994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6943362899867543e-06,
+      "loss": 0.6941,
+      "step": 4995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6913845214503065e-06,
+      "loss": 0.6956,
+      "step": 4996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6884350888000866e-06,
+      "loss": 0.6294,
+      "step": 4997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.685487992865299e-06,
+      "loss": 0.5686,
+      "step": 4998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.682543234474494e-06,
+      "loss": 0.6445,
+      "step": 4999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6796008144555586e-06,
+      "loss": 0.7358,
+      "step": 5000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6766607336357232e-06,
+      "loss": 0.7063,
+      "step": 5001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6737229928415676e-06,
+      "loss": 0.7092,
+      "step": 5002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 0.5898,
+      "step": 5003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6678545346332996e-06,
+      "loss": 0.6924,
+      "step": 5004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649238188690454e-06,
+      "loss": 0.5688,
+      "step": 5005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6619954464301902e-06,
+      "loss": 0.771,
+      "step": 5006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6590694181400136e-06,
+      "loss": 0.5715,
+      "step": 5007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656145734821143e-06,
+      "loss": 0.613,
+      "step": 5008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6532243972955397e-06,
+      "loss": 0.8398,
+      "step": 5009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6503054063845125e-06,
+      "loss": 0.6858,
+      "step": 5010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.647388762908707e-06,
+      "loss": 0.6619,
+      "step": 5011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6444744676881074e-06,
+      "loss": 0.5918,
+      "step": 5012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6415625215420416e-06,
+      "loss": 0.6553,
+      "step": 5013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6386529252891725e-06,
+      "loss": 0.7075,
+      "step": 5014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6357456797475069e-06,
+      "loss": 0.8828,
+      "step": 5015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6328407857343853e-06,
+      "loss": 0.6748,
+      "step": 5016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299382440664958e-06,
+      "loss": 0.6833,
+      "step": 5017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6270380555598531e-06,
+      "loss": 0.7034,
+      "step": 5018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6241402210298219e-06,
+      "loss": 0.5984,
+      "step": 5019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.621244741291097e-06,
+      "loss": 0.6497,
+      "step": 5020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6183516171577152e-06,
+      "loss": 0.7549,
+      "step": 5021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6154608494430512e-06,
+      "loss": 0.7759,
+      "step": 5022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6125724389598175e-06,
+      "loss": 0.834,
+      "step": 5023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6096863865200606e-06,
+      "loss": 0.7563,
+      "step": 5024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6068026929351644e-06,
+      "loss": 0.6377,
+      "step": 5025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6039213590158542e-06,
+      "loss": 0.6482,
+      "step": 5026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6010423855721857e-06,
+      "loss": 0.6228,
+      "step": 5027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598165773413558e-06,
+      "loss": 0.7825,
+      "step": 5028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5952915233486987e-06,
+      "loss": 0.656,
+      "step": 5029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5924196361856792e-06,
+      "loss": 0.6301,
+      "step": 5030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589550112731899e-06,
+      "loss": 0.6235,
+      "step": 5031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5866829537940975e-06,
+      "loss": 0.689,
+      "step": 5032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5838181601783519e-06,
+      "loss": 0.6807,
+      "step": 5033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.580955732690065e-06,
+      "loss": 0.5117,
+      "step": 5034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5780956721339847e-06,
+      "loss": 0.6335,
+      "step": 5035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5752379793141893e-06,
+      "loss": 0.752,
+      "step": 5036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5723826550340904e-06,
+      "loss": 0.793,
+      "step": 5037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5695297000964327e-06,
+      "loss": 0.7192,
+      "step": 5038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566679115303299e-06,
+      "loss": 0.7615,
+      "step": 5039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5638309014561015e-06,
+      "loss": 0.5859,
+      "step": 5040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5609850593555909e-06,
+      "loss": 0.6643,
+      "step": 5041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5581415898018437e-06,
+      "loss": 0.7659,
+      "step": 5042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5553004935942772e-06,
+      "loss": 0.8525,
+      "step": 5043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5524617715316348e-06,
+      "loss": 0.7063,
+      "step": 5044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549625424411998e-06,
+      "loss": 0.7139,
+      "step": 5045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5467914530327765e-06,
+      "loss": 0.6714,
+      "step": 5046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5439598581907167e-06,
+      "loss": 0.7788,
+      "step": 5047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5411306406818926e-06,
+      "loss": 0.7253,
+      "step": 5048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538303801301707e-06,
+      "loss": 0.6616,
+      "step": 5049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5354793408449043e-06,
+      "loss": 0.6753,
+      "step": 5050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532657260105549e-06,
+      "loss": 0.7209,
+      "step": 5051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5298375598770464e-06,
+      "loss": 0.6526,
+      "step": 5052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527020240952123e-06,
+      "loss": 0.7661,
+      "step": 5053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5242053041228445e-06,
+      "loss": 0.6083,
+      "step": 5054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213927501805992e-06,
+      "loss": 0.7307,
+      "step": 5055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5185825799161114e-06,
+      "loss": 0.7593,
+      "step": 5056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5157747941194346e-06,
+      "loss": 0.7031,
+      "step": 5057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.512969393579946e-06,
+      "loss": 0.7285,
+      "step": 5058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5101663790863597e-06,
+      "loss": 0.8911,
+      "step": 5059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5073657514267182e-06,
+      "loss": 0.7515,
+      "step": 5060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5045675113883872e-06,
+      "loss": 0.7004,
+      "step": 5061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5017716597580633e-06,
+      "loss": 0.7185,
+      "step": 5062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4989781973217776e-06,
+      "loss": 0.6077,
+      "step": 5063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4961871248648785e-06,
+      "loss": 0.7812,
+      "step": 5064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493398443172056e-06,
+      "loss": 0.6328,
+      "step": 5065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906121530273144e-06,
+      "loss": 0.9102,
+      "step": 5066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4878282552139977e-06,
+      "loss": 0.7627,
+      "step": 5067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4850467505147658e-06,
+      "loss": 0.7095,
+      "step": 5068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4822676397116153e-06,
+      "loss": 0.6135,
+      "step": 5069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4794909235858645e-06,
+      "loss": 0.6084,
+      "step": 5070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4767166029181635e-06,
+      "loss": 0.665,
+      "step": 5071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4739446784884837e-06,
+      "loss": 0.6904,
+      "step": 5072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4711751510761207e-06,
+      "loss": 0.6953,
+      "step": 5073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4684080214597064e-06,
+      "loss": 0.6633,
+      "step": 5074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4656432904171859e-06,
+      "loss": 0.645,
+      "step": 5075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462880958725843e-06,
+      "loss": 0.5574,
+      "step": 5076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4601210271622735e-06,
+      "loss": 0.5999,
+      "step": 5077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457363496502412e-06,
+      "loss": 0.7351,
+      "step": 5078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454608367521506e-06,
+      "loss": 0.6321,
+      "step": 5079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4518556409941342e-06,
+      "loss": 0.7302,
+      "step": 5080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491053176942004e-06,
+      "loss": 0.7312,
+      "step": 5081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 0.8057,
+      "step": 5082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4436118838688795e-06,
+      "loss": 0.7153,
+      "step": 5083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4408687748879157e-06,
+      "loss": 0.6951,
+      "step": 5084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381280722232404e-06,
+      "loss": 0.7158,
+      "step": 5085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4353897766453729e-06,
+      "loss": 0.7349,
+      "step": 5086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4326538889241615e-06,
+      "loss": 0.6548,
+      "step": 5087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4299204098287712e-06,
+      "loss": 0.6897,
+      "step": 5088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4271893401276972e-06,
+      "loss": 0.6851,
+      "step": 5089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244606805887474e-06,
+      "loss": 0.5081,
+      "step": 5090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.421734431979065e-06,
+      "loss": 0.6095,
+      "step": 5091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4190105950651e-06,
+      "loss": 0.7327,
+      "step": 5092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4162891706126392e-06,
+      "loss": 0.7043,
+      "step": 5093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4135701593867802e-06,
+      "loss": 0.6689,
+      "step": 5094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4108535621519514e-06,
+      "loss": 0.6603,
+      "step": 5095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4081393796718957e-06,
+      "loss": 0.629,
+      "step": 5096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4054276127096744e-06,
+      "loss": 0.7041,
+      "step": 5097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4027182620276814e-06,
+      "loss": 0.7009,
+      "step": 5098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4000113283876183e-06,
+      "loss": 0.6323,
+      "step": 5099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3973068125505186e-06,
+      "loss": 0.793,
+      "step": 5100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.394604715276724e-06,
+      "loss": 0.8262,
+      "step": 5101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3919050373259092e-06,
+      "loss": 0.7104,
+      "step": 5102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.389207779457058e-06,
+      "loss": 0.7031,
+      "step": 5103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3865129424284785e-06,
+      "loss": 0.6672,
+      "step": 5104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3838205269978e-06,
+      "loss": 0.6411,
+      "step": 5105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3811305339219705e-06,
+      "loss": 0.7512,
+      "step": 5106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3784429639572494e-06,
+      "loss": 0.6636,
+      "step": 5107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3757578178592268e-06,
+      "loss": 0.6157,
+      "step": 5108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3730750963828033e-06,
+      "loss": 0.752,
+      "step": 5109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3703948002821954e-06,
+      "loss": 0.9001,
+      "step": 5110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3677169303109494e-06,
+      "loss": 0.6992,
+      "step": 5111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650414872219174e-06,
+      "loss": 0.8398,
+      "step": 5112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.362368471767277e-06,
+      "loss": 0.7129,
+      "step": 5113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3596978846985176e-06,
+      "loss": 0.8245,
+      "step": 5114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3570297267664512e-06,
+      "loss": 0.7588,
+      "step": 5115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3543639987212032e-06,
+      "loss": 0.7881,
+      "step": 5116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3517007013122153e-06,
+      "loss": 0.6602,
+      "step": 5117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3490398352882505e-06,
+      "loss": 0.5609,
+      "step": 5118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3463814013973875e-06,
+      "loss": 0.6807,
+      "step": 5119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3437254003870148e-06,
+      "loss": 0.7288,
+      "step": 5120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3410718330038408e-06,
+      "loss": 0.5393,
+      "step": 5121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3384206999938943e-06,
+      "loss": 0.6377,
+      "step": 5122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3357720021025101e-06,
+      "loss": 0.681,
+      "step": 5123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3331257400743491e-06,
+      "loss": 0.7925,
+      "step": 5124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3304819146533765e-06,
+      "loss": 0.6,
+      "step": 5125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3278405265828843e-06,
+      "loss": 0.7195,
+      "step": 5126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325201576605467e-06,
+      "loss": 0.6655,
+      "step": 5127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322565065463043e-06,
+      "loss": 0.8145,
+      "step": 5128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3199309938968408e-06,
+      "loss": 0.728,
+      "step": 5129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172993626474062e-06,
+      "loss": 0.6806,
+      "step": 5130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.314670172454593e-06,
+      "loss": 0.6729,
+      "step": 5131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3120434240575763e-06,
+      "loss": 0.6367,
+      "step": 5132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309419118194838e-06,
+      "loss": 0.6063,
+      "step": 5133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067972556041753e-06,
+      "loss": 0.6455,
+      "step": 5134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304177837022702e-06,
+      "loss": 0.6211,
+      "step": 5135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3015608631868404e-06,
+      "loss": 0.6501,
+      "step": 5136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2989463348323284e-06,
+      "loss": 0.7708,
+      "step": 5137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2963342526942125e-06,
+      "loss": 0.5667,
+      "step": 5138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937246175068563e-06,
+      "loss": 0.7139,
+      "step": 5139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2911174300039342e-06,
+      "loss": 0.6902,
+      "step": 5140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885126909184286e-06,
+      "loss": 0.7612,
+      "step": 5141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2859104009826383e-06,
+      "loss": 0.7229,
+      "step": 5142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.283310560928175e-06,
+      "loss": 0.7461,
+      "step": 5143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2807131714859533e-06,
+      "loss": 0.7112,
+      "step": 5144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.278118233386204e-06,
+      "loss": 0.7422,
+      "step": 5145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2755257473584737e-06,
+      "loss": 0.6221,
+      "step": 5146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2729357141316078e-06,
+      "loss": 0.7495,
+      "step": 5147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2703481344337743e-06,
+      "loss": 0.6816,
+      "step": 5148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267763008992442e-06,
+      "loss": 0.6738,
+      "step": 5149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651803385343986e-06,
+      "loss": 0.7593,
+      "step": 5150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26260012378573e-06,
+      "loss": 0.7034,
+      "step": 5151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600223654718434e-06,
+      "loss": 0.6804,
+      "step": 5152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2574470643174496e-06,
+      "loss": 0.6377,
+      "step": 5153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2548742210465703e-06,
+      "loss": 0.6506,
+      "step": 5154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2523038363825324e-06,
+      "loss": 0.655,
+      "step": 5155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2497359110479779e-06,
+      "loss": 0.625,
+      "step": 5156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471704457648525e-06,
+      "loss": 0.7241,
+      "step": 5157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2446074412544084e-06,
+      "loss": 0.6716,
+      "step": 5158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2420468982372158e-06,
+      "loss": 0.6345,
+      "step": 5159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23948881743314e-06,
+      "loss": 0.7419,
+      "step": 5160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.719,
+      "step": 5161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2343800453403765e-06,
+      "loss": 0.6396,
+      "step": 5162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2318293554879669e-06,
+      "loss": 0.7427,
+      "step": 5163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2292811307212415e-06,
+      "loss": 0.7568,
+      "step": 5164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267353717566055e-06,
+      "loss": 0.6968,
+      "step": 5165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.224192079309775e-06,
+      "loss": 0.7747,
+      "step": 5166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2216512540957758e-06,
+      "loss": 0.6509,
+      "step": 5167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2191128968289324e-06,
+      "loss": 0.6078,
+      "step": 5168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2165770082228768e-06,
+      "loss": 0.7329,
+      "step": 5169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2140435889905545e-06,
+      "loss": 0.6497,
+      "step": 5170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2115126398442078e-06,
+      "loss": 0.7363,
+      "step": 5171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2089841614953913e-06,
+      "loss": 0.7217,
+      "step": 5172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2064581546549592e-06,
+      "loss": 0.6836,
+      "step": 5173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203934620033077e-06,
+      "loss": 0.7471,
+      "step": 5174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2014135583392072e-06,
+      "loss": 0.6592,
+      "step": 5175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1988949702821251e-06,
+      "loss": 0.7498,
+      "step": 5176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1963788565699075e-06,
+      "loss": 0.7412,
+      "step": 5177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1938652179099354e-06,
+      "loss": 0.6157,
+      "step": 5178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1913540550088899e-06,
+      "loss": 0.5822,
+      "step": 5179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1888453685727664e-06,
+      "loss": 0.6497,
+      "step": 5180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186339159306853e-06,
+      "loss": 0.6165,
+      "step": 5181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183835427915745e-06,
+      "loss": 0.7471,
+      "step": 5182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1813341751033459e-06,
+      "loss": 0.728,
+      "step": 5183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1788354015728543e-06,
+      "loss": 0.5752,
+      "step": 5184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1763391080267815e-06,
+      "loss": 0.7085,
+      "step": 5185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1738452951669311e-06,
+      "loss": 0.5718,
+      "step": 5186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713539636944149e-06,
+      "loss": 0.6958,
+      "step": 5187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1688651143096486e-06,
+      "loss": 0.5459,
+      "step": 5188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1663787477123488e-06,
+      "loss": 0.8936,
+      "step": 5189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1638948646015291e-06,
+      "loss": 0.6274,
+      "step": 5190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161413465675514e-06,
+      "loss": 0.6973,
+      "step": 5191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1589345516319216e-06,
+      "loss": 0.7354,
+      "step": 5192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156458123167674e-06,
+      "loss": 0.7402,
+      "step": 5193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1539841809789965e-06,
+      "loss": 0.6697,
+      "step": 5194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151512725761411e-06,
+      "loss": 0.6136,
+      "step": 5195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.149043758209748e-06,
+      "loss": 0.8008,
+      "step": 5196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1465772790181284e-06,
+      "loss": 0.7668,
+      "step": 5197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1441132888799822e-06,
+      "loss": 0.8159,
+      "step": 5198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141651788488034e-06,
+      "loss": 0.6479,
+      "step": 5199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1391927785343114e-06,
+      "loss": 0.571,
+      "step": 5200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1367362597101429e-06,
+      "loss": 0.748,
+      "step": 5201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1342822327061532e-06,
+      "loss": 0.71,
+      "step": 5202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1318306982122674e-06,
+      "loss": 0.7256,
+      "step": 5203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293816569177129e-06,
+      "loss": 0.679,
+      "step": 5204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1269351095110126e-06,
+      "loss": 0.8193,
+      "step": 5205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244910566799872e-06,
+      "loss": 0.6277,
+      "step": 5206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1220494991117614e-06,
+      "loss": 0.6685,
+      "step": 5207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.119610437492753e-06,
+      "loss": 0.6914,
+      "step": 5208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1171738725086833e-06,
+      "loss": 0.6726,
+      "step": 5209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1147398048445646e-06,
+      "loss": 0.7578,
+      "step": 5210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1123082351847136e-06,
+      "loss": 0.6887,
+      "step": 5211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1098791642127427e-06,
+      "loss": 0.8135,
+      "step": 5212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107452592611563e-06,
+      "loss": 0.6343,
+      "step": 5213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050285210633782e-06,
+      "loss": 0.5636,
+      "step": 5214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.102606950249696e-06,
+      "loss": 0.7573,
+      "step": 5215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1001878808513123e-06,
+      "loss": 0.7803,
+      "step": 5216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0977713135483313e-06,
+      "loss": 0.6206,
+      "step": 5217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0953572490201426e-06,
+      "loss": 0.4734,
+      "step": 5218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092945687945437e-06,
+      "loss": 0.7405,
+      "step": 5219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0905366310022048e-06,
+      "loss": 0.5815,
+      "step": 5220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0881300788677241e-06,
+      "loss": 0.6948,
+      "step": 5221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0857260322185769e-06,
+      "loss": 0.7644,
+      "step": 5222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0833244917306384e-06,
+      "loss": 0.5811,
+      "step": 5223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.080925458079074e-06,
+      "loss": 0.8423,
+      "step": 5224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078528931938352e-06,
+      "loss": 0.7051,
+      "step": 5225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076134913982233e-06,
+      "loss": 0.709,
+      "step": 5226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073743404883769e-06,
+      "loss": 0.6469,
+      "step": 5227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0713544053153124e-06,
+      "loss": 0.7273,
+      "step": 5228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0689679159485056e-06,
+      "loss": 0.6663,
+      "step": 5229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0665839374542853e-06,
+      "loss": 0.6899,
+      "step": 5230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0642024705028875e-06,
+      "loss": 0.6572,
+      "step": 5231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0618235157638336e-06,
+      "loss": 0.7622,
+      "step": 5232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.059447073905948e-06,
+      "loss": 0.8354,
+      "step": 5233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0570731455973415e-06,
+      "loss": 0.6218,
+      "step": 5234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0547017315054208e-06,
+      "loss": 0.6805,
+      "step": 5235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0523328322968863e-06,
+      "loss": 0.6646,
+      "step": 5236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0499664486377349e-06,
+      "loss": 0.7769,
+      "step": 5237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476025811932456e-06,
+      "loss": 0.6682,
+      "step": 5238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0452412306280023e-06,
+      "loss": 0.6277,
+      "step": 5239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 0.6887,
+      "step": 5240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0405260827900187e-06,
+      "loss": 0.608,
+      "step": 5241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038172286842899e-06,
+      "loss": 0.6638,
+      "step": 5242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0358210104262557e-06,
+      "loss": 0.6836,
+      "step": 5243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033472254201132e-06,
+      "loss": 0.6528,
+      "step": 5244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0311260188278537e-06,
+      "loss": 0.8538,
+      "step": 5245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.028782304966044e-06,
+      "loss": 0.8569,
+      "step": 5246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0264411132746165e-06,
+      "loss": 0.8057,
+      "step": 5247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.024102444411771e-06,
+      "loss": 0.6772,
+      "step": 5248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021766299035003e-06,
+      "loss": 0.7549,
+      "step": 5249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0194326778010998e-06,
+      "loss": 0.635,
+      "step": 5250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0171015813661322e-06,
+      "loss": 0.8328,
+      "step": 5251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147730103854681e-06,
+      "loss": 0.6079,
+      "step": 5252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0124469655137626e-06,
+      "loss": 0.7266,
+      "step": 5253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010123447404956e-06,
+      "loss": 0.6421,
+      "step": 5254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007802456712288e-06,
+      "loss": 0.5989,
+      "step": 5255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0054839940882787e-06,
+      "loss": 0.6655,
+      "step": 5256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031680601847448e-06,
+      "loss": 0.6604,
+      "step": 5257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0008546556527854e-06,
+      "loss": 0.7856,
+      "step": 5258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.985437811427934e-07,
+      "loss": 0.7437,
+      "step": 5259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.96235437304447e-07,
+      "loss": 0.8369,
+      "step": 5260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.939296247867191e-07,
+      "loss": 0.7588,
+      "step": 5261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.916263442378604e-07,
+      "loss": 0.697,
+      "step": 5262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.893255963054205e-07,
+      "loss": 0.6926,
+      "step": 5263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870273816362286e-07,
+      "loss": 0.6082,
+      "step": 5264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84731700876408e-07,
+      "loss": 0.7251,
+      "step": 5265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.824385546713666e-07,
+      "loss": 0.5984,
+      "step": 5266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.801479436657979e-07,
+      "loss": 0.7361,
+      "step": 5267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.778598685036878e-07,
+      "loss": 0.7524,
+      "step": 5268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.755743298283027e-07,
+      "loss": 0.8184,
+      "step": 5269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73291328282202e-07,
+      "loss": 0.8037,
+      "step": 5270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71010864507229e-07,
+      "loss": 0.5801,
+      "step": 5271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.687329391445166e-07,
+      "loss": 0.7302,
+      "step": 5272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.664575528344754e-07,
+      "loss": 0.7554,
+      "step": 5273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641847062168142e-07,
+      "loss": 0.7607,
+      "step": 5274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61914399930517e-07,
+      "loss": 0.752,
+      "step": 5275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.596466346138611e-07,
+      "loss": 0.6143,
+      "step": 5276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.573814109044055e-07,
+      "loss": 0.7275,
+      "step": 5277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551187294389941e-07,
+      "loss": 0.636,
+      "step": 5278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.528585908537613e-07,
+      "loss": 0.6838,
+      "step": 5279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.506009957841201e-07,
+      "loss": 0.7261,
+      "step": 5280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483459448647736e-07,
+      "loss": 0.5859,
+      "step": 5281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.460934387297038e-07,
+      "loss": 0.6011,
+      "step": 5282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.438434780121841e-07,
+      "loss": 0.7808,
+      "step": 5283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.415960633447674e-07,
+      "loss": 0.6062,
+      "step": 5284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393511953592949e-07,
+      "loss": 0.7368,
+      "step": 5285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.371088746868861e-07,
+      "loss": 0.6748,
+      "step": 5286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.348691019579515e-07,
+      "loss": 0.7104,
+      "step": 5287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326318778021759e-07,
+      "loss": 0.726,
+      "step": 5288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.303972028485386e-07,
+      "loss": 0.7881,
+      "step": 5289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.281650777252938e-07,
+      "loss": 0.8438,
+      "step": 5290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.259355030599815e-07,
+      "loss": 0.7454,
+      "step": 5291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.23708479479426e-07,
+      "loss": 0.5356,
+      "step": 5292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.214840076097309e-07,
+      "loss": 0.7515,
+      "step": 5293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.19262088076287e-07,
+      "loss": 0.7336,
+      "step": 5294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.170427215037647e-07,
+      "loss": 0.7161,
+      "step": 5295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.148259085161204e-07,
+      "loss": 0.7544,
+      "step": 5296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.126116497365833e-07,
+      "loss": 0.7144,
+      "step": 5297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.103999457876767e-07,
+      "loss": 0.7456,
+      "step": 5298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.081907972911941e-07,
+      "loss": 0.463,
+      "step": 5299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.05984204868221e-07,
+      "loss": 0.6711,
+      "step": 5300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.037801691391168e-07,
+      "loss": 0.6167,
+      "step": 5301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.015786907235235e-07,
+      "loss": 0.6576,
+      "step": 5302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993797702403695e-07,
+      "loss": 0.6567,
+      "step": 5303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.971834083078534e-07,
+      "loss": 0.7681,
+      "step": 5304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.949896055434682e-07,
+      "loss": 0.7268,
+      "step": 5305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927983625639735e-07,
+      "loss": 0.7627,
+      "step": 5306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.906096799854203e-07,
+      "loss": 0.8008,
+      "step": 5307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.884235584231326e-07,
+      "loss": 0.792,
+      "step": 5308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.862399984917214e-07,
+      "loss": 0.7798,
+      "step": 5309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.840590008050687e-07,
+      "loss": 0.7158,
+      "step": 5310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818805659763441e-07,
+      "loss": 0.6224,
+      "step": 5311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7970469461799e-07,
+      "loss": 0.621,
+      "step": 5312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.775313873417346e-07,
+      "loss": 0.7441,
+      "step": 5313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.753606447585816e-07,
+      "loss": 0.6729,
+      "step": 5314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.731924674788106e-07,
+      "loss": 0.7456,
+      "step": 5315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710268561119872e-07,
+      "loss": 0.7024,
+      "step": 5316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.6886381126695e-07,
+      "loss": 0.731,
+      "step": 5317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.667033335518183e-07,
+      "loss": 0.7734,
+      "step": 5318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.8057,
+      "step": 5319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623900819401421e-07,
+      "loss": 0.7078,
+      "step": 5320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.602373092562244e-07,
+      "loss": 0.6704,
+      "step": 5321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58087106127471e-07,
+      "loss": 0.6504,
+      "step": 5322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.55939473158387e-07,
+      "loss": 0.6946,
+      "step": 5323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537944109527629e-07,
+      "loss": 0.772,
+      "step": 5324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.516519201136598e-07,
+      "loss": 0.7202,
+      "step": 5325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.495120012434155e-07,
+      "loss": 0.7197,
+      "step": 5326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.473746549436523e-07,
+      "loss": 0.6877,
+      "step": 5327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452398818152596e-07,
+      "loss": 0.667,
+      "step": 5328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.431076824584106e-07,
+      "loss": 0.8262,
+      "step": 5329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409780574725534e-07,
+      "loss": 0.7004,
+      "step": 5330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388510074564082e-07,
+      "loss": 0.6179,
+      "step": 5331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367265330079766e-07,
+      "loss": 0.6729,
+      "step": 5332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346046347245351e-07,
+      "loss": 0.6951,
+      "step": 5333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.3248531320263e-07,
+      "loss": 0.8511,
+      "step": 5334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.303685690380935e-07,
+      "loss": 0.7375,
+      "step": 5335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.282544028260231e-07,
+      "loss": 0.571,
+      "step": 5336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.261428151607998e-07,
+      "loss": 0.585,
+      "step": 5337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240338066360742e-07,
+      "loss": 0.6111,
+      "step": 5338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.219273778447712e-07,
+      "loss": 0.7502,
+      "step": 5339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198235293790957e-07,
+      "loss": 0.7495,
+      "step": 5340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.177222618305225e-07,
+      "loss": 0.5289,
+      "step": 5341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.156235757898024e-07,
+      "loss": 0.6384,
+      "step": 5342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.135274718469622e-07,
+      "loss": 0.6992,
+      "step": 5343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.114339505913016e-07,
+      "loss": 0.6814,
+      "step": 5344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.0934301261139e-07,
+      "loss": 0.6062,
+      "step": 5345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.07254658495078e-07,
+      "loss": 0.74,
+      "step": 5346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.05168888829484e-07,
+      "loss": 0.6577,
+      "step": 5347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.030857042010031e-07,
+      "loss": 0.6135,
+      "step": 5348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.010051051953027e-07,
+      "loss": 0.8311,
+      "step": 5349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.989270923973191e-07,
+      "loss": 0.8433,
+      "step": 5350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.968516663912696e-07,
+      "loss": 0.7581,
+      "step": 5351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.947788277606372e-07,
+      "loss": 0.7029,
+      "step": 5352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.927085770881815e-07,
+      "loss": 0.7173,
+      "step": 5353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906409149559347e-07,
+      "loss": 0.7561,
+      "step": 5354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.885758419451961e-07,
+      "loss": 0.8127,
+      "step": 5355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.865133586365436e-07,
+      "loss": 0.6674,
+      "step": 5356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.844534656098257e-07,
+      "loss": 0.7151,
+      "step": 5357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823961634441579e-07,
+      "loss": 0.7476,
+      "step": 5358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.803414527179343e-07,
+      "loss": 0.8145,
+      "step": 5359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.782893340088116e-07,
+      "loss": 0.6311,
+      "step": 5360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.762398078937283e-07,
+      "loss": 0.8245,
+      "step": 5361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.74192874948887e-07,
+      "loss": 0.7974,
+      "step": 5362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.721485357497593e-07,
+      "loss": 0.7944,
+      "step": 5363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.701067908710968e-07,
+      "loss": 0.7449,
+      "step": 5364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.680676408869114e-07,
+      "loss": 0.6306,
+      "step": 5365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.660310863704923e-07,
+      "loss": 0.6448,
+      "step": 5366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.63997127894398e-07,
+      "loss": 0.6816,
+      "step": 5367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.619657660304558e-07,
+      "loss": 0.6648,
+      "step": 5368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.599370013497609e-07,
+      "loss": 0.8022,
+      "step": 5369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.579108344226849e-07,
+      "loss": 0.6619,
+      "step": 5370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.55887265818861e-07,
+      "loss": 0.77,
+      "step": 5371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.538662961071974e-07,
+      "loss": 0.7136,
+      "step": 5372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.518479258558708e-07,
+      "loss": 0.7065,
+      "step": 5373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.49832155632324e-07,
+      "loss": 0.7458,
+      "step": 5374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.478189860032736e-07,
+      "loss": 0.6897,
+      "step": 5375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.458084175347002e-07,
+      "loss": 0.7368,
+      "step": 5376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.438004507918562e-07,
+      "loss": 0.7549,
+      "step": 5377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.417950863392632e-07,
+      "loss": 0.7407,
+      "step": 5378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397923247407102e-07,
+      "loss": 0.7283,
+      "step": 5379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377921665592525e-07,
+      "loss": 0.8418,
+      "step": 5380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.357946123572158e-07,
+      "loss": 0.7371,
+      "step": 5381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.337996626961918e-07,
+      "loss": 0.6106,
+      "step": 5382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.318073181370444e-07,
+      "loss": 0.7856,
+      "step": 5383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298175792398976e-07,
+      "loss": 0.8035,
+      "step": 5384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.278304465641506e-07,
+      "loss": 0.6716,
+      "step": 5385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258459206684642e-07,
+      "loss": 0.6855,
+      "step": 5386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.238640021107679e-07,
+      "loss": 0.9619,
+      "step": 5387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.218846914482624e-07,
+      "loss": 0.8252,
+      "step": 5388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.199079892374072e-07,
+      "loss": 0.8433,
+      "step": 5389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179338960339343e-07,
+      "loss": 0.77,
+      "step": 5390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.159624123928422e-07,
+      "loss": 0.6936,
+      "step": 5391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.139935388683949e-07,
+      "loss": 0.8848,
+      "step": 5392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12027276014119e-07,
+      "loss": 0.634,
+      "step": 5393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10063624382813e-07,
+      "loss": 0.6792,
+      "step": 5394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081025845265355e-07,
+      "loss": 0.6294,
+      "step": 5395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.061441569966165e-07,
+      "loss": 0.6526,
+      "step": 5396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.041883423436469e-07,
+      "loss": 0.7898,
+      "step": 5397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.7864,
+      "step": 5398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.002845538672587e-07,
+      "loss": 0.8628,
+      "step": 5399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983365811413489e-07,
+      "loss": 0.7395,
+      "step": 5400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.963912234874137e-07,
+      "loss": 0.668,
+      "step": 5401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9444848145237e-07,
+      "loss": 0.7026,
+      "step": 5402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925083555824041e-07,
+      "loss": 0.7407,
+      "step": 5403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905708464229588e-07,
+      "loss": 0.6738,
+      "step": 5404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.886359545187493e-07,
+      "loss": 0.7876,
+      "step": 5405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.867036804137495e-07,
+      "loss": 0.5941,
+      "step": 5406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.847740246512014e-07,
+      "loss": 0.7764,
+      "step": 5407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.828469877736066e-07,
+      "loss": 0.717,
+      "step": 5408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809225703227352e-07,
+      "loss": 0.707,
+      "step": 5409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.790007728396175e-07,
+      "loss": 0.7537,
+      "step": 5410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770815958645449e-07,
+      "loss": 0.8496,
+      "step": 5411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.751650399370813e-07,
+      "loss": 0.7007,
+      "step": 5412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.732511055960412e-07,
+      "loss": 0.8911,
+      "step": 5413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.71339793379513e-07,
+      "loss": 0.7144,
+      "step": 5414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.694311038248424e-07,
+      "loss": 0.6793,
+      "step": 5415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.675250374686404e-07,
+      "loss": 0.8145,
+      "step": 5416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.656215948467759e-07,
+      "loss": 0.833,
+      "step": 5417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.637207764943854e-07,
+      "loss": 0.6021,
+      "step": 5418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.618225829458647e-07,
+      "loss": 0.7925,
+      "step": 5419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599270147348735e-07,
+      "loss": 0.6931,
+      "step": 5420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.58034072394329e-07,
+      "loss": 0.6968,
+      "step": 5421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.561437564564177e-07,
+      "loss": 0.6404,
+      "step": 5422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.542560674525822e-07,
+      "loss": 0.639,
+      "step": 5423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.523710059135257e-07,
+      "loss": 0.6863,
+      "step": 5424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504885723692156e-07,
+      "loss": 0.593,
+      "step": 5425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.486087673488817e-07,
+      "loss": 0.7231,
+      "step": 5426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.467315913810124e-07,
+      "loss": 0.7925,
+      "step": 5427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448570449933555e-07,
+      "loss": 0.7646,
+      "step": 5428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.429851287129241e-07,
+      "loss": 0.6753,
+      "step": 5429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.411158430659869e-07,
+      "loss": 0.5908,
+      "step": 5430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.392491885780772e-07,
+      "loss": 0.7559,
+      "step": 5431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.373851657739838e-07,
+      "loss": 0.6597,
+      "step": 5432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355237751777621e-07,
+      "loss": 0.7419,
+      "step": 5433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.336650173127224e-07,
+      "loss": 0.6365,
+      "step": 5434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318088927014354e-07,
+      "loss": 0.7358,
+      "step": 5435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.299554018657328e-07,
+      "loss": 0.6294,
+      "step": 5436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.281045453267076e-07,
+      "loss": 0.7783,
+      "step": 5437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262563236047059e-07,
+      "loss": 0.6665,
+      "step": 5438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.244107372193409e-07,
+      "loss": 0.625,
+      "step": 5439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225677866894808e-07,
+      "loss": 0.6067,
+      "step": 5440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207274725332524e-07,
+      "loss": 0.614,
+      "step": 5441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.188897952680417e-07,
+      "loss": 0.8132,
+      "step": 5442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170547554104933e-07,
+      "loss": 0.6926,
+      "step": 5443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152223534765134e-07,
+      "loss": 0.6482,
+      "step": 5444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.133925899812609e-07,
+      "loss": 0.6536,
+      "step": 5445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.115654654391579e-07,
+      "loss": 0.6191,
+      "step": 5446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097409803638832e-07,
+      "loss": 0.6809,
+      "step": 5447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.079191352683688e-07,
+      "loss": 0.6917,
+      "step": 5448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.060999306648129e-07,
+      "loss": 0.7397,
+      "step": 5449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042833670646652e-07,
+      "loss": 0.7238,
+      "step": 5450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.024694449786373e-07,
+      "loss": 0.7744,
+      "step": 5451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006581649166921e-07,
+      "loss": 0.6323,
+      "step": 5452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988495273880568e-07,
+      "loss": 0.6875,
+      "step": 5453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970435329012092e-07,
+      "loss": 0.5225,
+      "step": 5454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952401819638887e-07,
+      "loss": 0.624,
+      "step": 5455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.934394750830896e-07,
+      "loss": 0.7417,
+      "step": 5456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.916414127650638e-07,
+      "loss": 0.8171,
+      "step": 5457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.898459955153179e-07,
+      "loss": 0.6368,
+      "step": 5458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.880532238386161e-07,
+      "loss": 0.8445,
+      "step": 5459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862630982389772e-07,
+      "loss": 0.7236,
+      "step": 5460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844756192196798e-07,
+      "loss": 0.7808,
+      "step": 5461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826907872832576e-07,
+      "loss": 0.7158,
+      "step": 5462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.809086029314947e-07,
+      "loss": 0.7178,
+      "step": 5463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.791290666654392e-07,
+      "loss": 0.7048,
+      "step": 5464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.773521789853853e-07,
+      "loss": 0.873,
+      "step": 5465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755779403908923e-07,
+      "loss": 0.7239,
+      "step": 5466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.738063513807668e-07,
+      "loss": 0.6575,
+      "step": 5467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.720374124530759e-07,
+      "loss": 0.7876,
+      "step": 5468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.702711241051384e-07,
+      "loss": 0.7102,
+      "step": 5469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.685074868335294e-07,
+      "loss": 0.6262,
+      "step": 5470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6674650113408e-07,
+      "loss": 0.7217,
+      "step": 5471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.649881675018698e-07,
+      "loss": 0.6782,
+      "step": 5472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6323248643124e-07,
+      "loss": 0.7874,
+      "step": 5473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.614794584157834e-07,
+      "loss": 0.698,
+      "step": 5474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.597290839483471e-07,
+      "loss": 0.7134,
+      "step": 5475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579813635210308e-07,
+      "loss": 0.7632,
+      "step": 5476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 0.8174,
+      "step": 5477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.544938867514305e-07,
+      "loss": 0.7026,
+      "step": 5478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.527541313896179e-07,
+      "loss": 0.7461,
+      "step": 5479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.510170320288644e-07,
+      "loss": 0.6931,
+      "step": 5480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.492825891575404e-07,
+      "loss": 0.6924,
+      "step": 5481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.475508032632671e-07,
+      "loss": 0.5988,
+      "step": 5482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458216748329182e-07,
+      "loss": 0.5825,
+      "step": 5483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.440952043526215e-07,
+      "loss": 0.6831,
+      "step": 5484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.423713923077579e-07,
+      "loss": 0.6987,
+      "step": 5485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.406502391829615e-07,
+      "loss": 0.7432,
+      "step": 5486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389317454621146e-07,
+      "loss": 0.7788,
+      "step": 5487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.372159116283592e-07,
+      "loss": 0.7139,
+      "step": 5488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.35502738164082e-07,
+      "loss": 0.6082,
+      "step": 5489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.337922255509265e-07,
+      "loss": 0.5659,
+      "step": 5490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.320843742697846e-07,
+      "loss": 0.8052,
+      "step": 5491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303791848008055e-07,
+      "loss": 0.6982,
+      "step": 5492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286766576233837e-07,
+      "loss": 0.6401,
+      "step": 5493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269767932161696e-07,
+      "loss": 0.6321,
+      "step": 5494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.25279592057063e-07,
+      "loss": 0.6636,
+      "step": 5495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.235850546232146e-07,
+      "loss": 0.7527,
+      "step": 5496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.218931813910278e-07,
+      "loss": 0.7004,
+      "step": 5497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.202039728361563e-07,
+      "loss": 0.7632,
+      "step": 5498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.185174294335061e-07,
+      "loss": 0.5941,
+      "step": 5499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168335516572287e-07,
+      "loss": 0.6929,
+      "step": 5500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151523399807345e-07,
+      "loss": 0.811,
+      "step": 5501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134737948766766e-07,
+      "loss": 0.7068,
+      "step": 5502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.117979168169629e-07,
+      "loss": 0.7432,
+      "step": 5503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101247062727488e-07,
+      "loss": 0.6616,
+      "step": 5504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084541637144435e-07,
+      "loss": 0.7339,
+      "step": 5505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067862896117026e-07,
+      "loss": 0.7573,
+      "step": 5506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051210844334309e-07,
+      "loss": 0.7871,
+      "step": 5507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03458548647785e-07,
+      "loss": 0.6375,
+      "step": 5508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.017986827221733e-07,
+      "loss": 0.896,
+      "step": 5509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001414871232502e-07,
+      "loss": 0.6172,
+      "step": 5510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984869623169175e-07,
+      "loss": 0.771,
+      "step": 5511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968351087683321e-07,
+      "loss": 0.7014,
+      "step": 5512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951859269418935e-07,
+      "loss": 0.6865,
+      "step": 5513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935394173012553e-07,
+      "loss": 0.5869,
+      "step": 5514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.918955803093162e-07,
+      "loss": 0.6411,
+      "step": 5515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.902544164282263e-07,
+      "loss": 0.7437,
+      "step": 5516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886159261193801e-07,
+      "loss": 0.6533,
+      "step": 5517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.869801098434279e-07,
+      "loss": 0.7554,
+      "step": 5518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853469680602585e-07,
+      "loss": 0.7236,
+      "step": 5519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83716501229018e-07,
+      "loss": 0.6562,
+      "step": 5520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820887098080918e-07,
+      "loss": 0.8237,
+      "step": 5521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.804635942551206e-07,
+      "loss": 0.7002,
+      "step": 5522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788411550269911e-07,
+      "loss": 0.6846,
+      "step": 5523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772213925798331e-07,
+      "loss": 0.7485,
+      "step": 5524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7560430736902865e-07,
+      "loss": 0.7827,
+      "step": 5525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.739898998492043e-07,
+      "loss": 0.647,
+      "step": 5526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7237817047423694e-07,
+      "loss": 0.5977,
+      "step": 5527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7076911969724635e-07,
+      "loss": 0.8052,
+      "step": 5528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.69162747970604e-07,
+      "loss": 0.6917,
+      "step": 5529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.675590557459253e-07,
+      "loss": 0.6956,
+      "step": 5530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.659580434740685e-07,
+      "loss": 0.8082,
+      "step": 5531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6435971160514683e-07,
+      "loss": 0.8667,
+      "step": 5532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627640605885142e-07,
+      "loss": 0.8345,
+      "step": 5533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61171090872774e-07,
+      "loss": 0.7285,
+      "step": 5534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5958080290577133e-07,
+      "loss": 0.7063,
+      "step": 5535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.57993197134603e-07,
+      "loss": 0.719,
+      "step": 5536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.564082740056064e-07,
+      "loss": 0.812,
+      "step": 5537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548260339643695e-07,
+      "loss": 0.6597,
+      "step": 5538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5324647745572103e-07,
+      "loss": 0.7729,
+      "step": 5539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.516696049237412e-07,
+      "loss": 0.7808,
+      "step": 5540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5009541681174993e-07,
+      "loss": 0.7163,
+      "step": 5541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4852391356231537e-07,
+      "loss": 0.8557,
+      "step": 5542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.469550956172497e-07,
+      "loss": 0.8892,
+      "step": 5543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.453889634176134e-07,
+      "loss": 0.7334,
+      "step": 5544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.438255174037054e-07,
+      "loss": 0.7773,
+      "step": 5545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4226475801507517e-07,
+      "loss": 0.6431,
+      "step": 5546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.407066856905162e-07,
+      "loss": 0.769,
+      "step": 5547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391513008680637e-07,
+      "loss": 0.6631,
+      "step": 5548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375986039850011e-07,
+      "loss": 0.6777,
+      "step": 5549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.360485954778504e-07,
+      "loss": 0.6809,
+      "step": 5550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.345012757823841e-07,
+      "loss": 0.7178,
+      "step": 5551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3295664533361535e-07,
+      "loss": 0.7112,
+      "step": 5552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.314147045658024e-07,
+      "loss": 0.6309,
+      "step": 5553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2987545391244613e-07,
+      "loss": 0.5856,
+      "step": 5554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2833889380629155e-07,
+      "loss": 0.7632,
+      "step": 5555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 0.676,
+      "step": 5556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2527384696278706e-07,
+      "loss": 0.718,
+      "step": 5557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.237453610871478e-07,
+      "loss": 0.9355,
+      "step": 5558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222195674821239e-07,
+      "loss": 0.729,
+      "step": 5559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2069646657668215e-07,
+      "loss": 0.6151,
+      "step": 5560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1917605879902457e-07,
+      "loss": 0.7415,
+      "step": 5561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.176583445766003e-07,
+      "loss": 0.75,
+      "step": 5562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1614332433609797e-07,
+      "loss": 0.7124,
+      "step": 5563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.146309985034547e-07,
+      "loss": 0.697,
+      "step": 5564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1312136750384147e-07,
+      "loss": 0.5823,
+      "step": 5565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.116144317616799e-07,
+      "loss": 0.714,
+      "step": 5566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101101917006278e-07,
+      "loss": 0.6624,
+      "step": 5567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.08608647743588e-07,
+      "loss": 0.8457,
+      "step": 5568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.071098003127072e-07,
+      "loss": 0.707,
+      "step": 5569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0561364982936944e-07,
+      "loss": 0.7329,
+      "step": 5570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.041201967142039e-07,
+      "loss": 0.7505,
+      "step": 5571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0262944138708017e-07,
+      "loss": 0.8508,
+      "step": 5572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0114138426710966e-07,
+      "loss": 0.634,
+      "step": 5573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.996560257726434e-07,
+      "loss": 0.7683,
+      "step": 5574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9817336632127833e-07,
+      "loss": 0.8701,
+      "step": 5575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9669340632984777e-07,
+      "loss": 0.7629,
+      "step": 5576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9521614621442884e-07,
+      "loss": 0.6211,
+      "step": 5577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.937415863903382e-07,
+      "loss": 0.8313,
+      "step": 5578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.922697272721354e-07,
+      "loss": 0.562,
+      "step": 5579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908005692736172e-07,
+      "loss": 0.7197,
+      "step": 5580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.893341128078232e-07,
+      "loss": 0.6165,
+      "step": 5581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8787035828703577e-07,
+      "loss": 0.6985,
+      "step": 5582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.864093061227725e-07,
+      "loss": 0.7998,
+      "step": 5583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8495095672579584e-07,
+      "loss": 0.7246,
+      "step": 5584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8349531050610456e-07,
+      "loss": 0.7271,
+      "step": 5585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8204236787294237e-07,
+      "loss": 0.6992,
+      "step": 5586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8059212923478693e-07,
+      "loss": 0.718,
+      "step": 5587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7914459499936106e-07,
+      "loss": 0.5748,
+      "step": 5588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.776997655736225e-07,
+      "loss": 0.6355,
+      "step": 5589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.762576413637731e-07,
+      "loss": 0.6985,
+      "step": 5590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.748182227752506e-07,
+      "loss": 0.825,
+      "step": 5591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.733815102127336e-07,
+      "loss": 0.605,
+      "step": 5592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.719475040801412e-07,
+      "loss": 0.6127,
+      "step": 5593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.705162047806277e-07,
+      "loss": 0.7822,
+      "step": 5594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690876127165932e-07,
+      "loss": 0.7507,
+      "step": 5595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676617282896666e-07,
+      "loss": 0.8047,
+      "step": 5596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6623855190072723e-07,
+      "loss": 0.7698,
+      "step": 5597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.648180839498827e-07,
+      "loss": 0.7139,
+      "step": 5598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.634003248364881e-07,
+      "loss": 0.6221,
+      "step": 5599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6198527495912795e-07,
+      "loss": 0.7278,
+      "step": 5600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6057293471563284e-07,
+      "loss": 0.6774,
+      "step": 5601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591633045030662e-07,
+      "loss": 0.7002,
+      "step": 5602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5775638471773454e-07,
+      "loss": 0.5942,
+      "step": 5603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5635217575517687e-07,
+      "loss": 0.73,
+      "step": 5604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5495067801017526e-07,
+      "loss": 0.7603,
+      "step": 5605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5355189187674556e-07,
+      "loss": 0.9058,
+      "step": 5606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5215581774814325e-07,
+      "loss": 0.7437,
+      "step": 5607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.507624560168621e-07,
+      "loss": 0.7031,
+      "step": 5608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.493718070746299e-07,
+      "loss": 0.6897,
+      "step": 5609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479838713124162e-07,
+      "loss": 0.7791,
+      "step": 5610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465986491204243e-07,
+      "loss": 0.6597,
+      "step": 5611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.452161408880961e-07,
+      "loss": 0.7317,
+      "step": 5612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.438363470041106e-07,
+      "loss": 0.8828,
+      "step": 5613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.424592678563843e-07,
+      "loss": 0.8203,
+      "step": 5614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410849038320674e-07,
+      "loss": 0.7368,
+      "step": 5615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397132553175497e-07,
+      "loss": 0.7627,
+      "step": 5616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.383443226984584e-07,
+      "loss": 0.5663,
+      "step": 5617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.369781063596533e-07,
+      "loss": 0.7478,
+      "step": 5618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3561460668523505e-07,
+      "loss": 0.6787,
+      "step": 5619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3425382405853577e-07,
+      "loss": 0.6506,
+      "step": 5620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3289575886212955e-07,
+      "loss": 0.7388,
+      "step": 5621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.315404114778198e-07,
+      "loss": 0.8413,
+      "step": 5622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301877822866506e-07,
+      "loss": 0.9507,
+      "step": 5623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2883787166890116e-07,
+      "loss": 0.7148,
+      "step": 5624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2749068000408445e-07,
+      "loss": 0.7598,
+      "step": 5625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2614620767095094e-07,
+      "loss": 0.739,
+      "step": 5626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2480445504748605e-07,
+      "loss": 0.731,
+      "step": 5627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.234654225109102e-07,
+      "loss": 0.9116,
+      "step": 5628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.221291104376789e-07,
+      "loss": 0.6794,
+      "step": 5629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2079551920348395e-07,
+      "loss": 0.7778,
+      "step": 5630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.194646491832509e-07,
+      "loss": 0.731,
+      "step": 5631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1813650075114275e-07,
+      "loss": 0.7871,
+      "step": 5632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.16811074280553e-07,
+      "loss": 0.7778,
+      "step": 5633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.154883701441136e-07,
+      "loss": 0.7651,
+      "step": 5634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.8643,
+      "step": 5635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1285113036038094e-07,
+      "loss": 0.8206,
+      "step": 5636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.115365954545213e-07,
+      "loss": 0.6846,
+      "step": 5637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1022478436568135e-07,
+      "loss": 0.6619,
+      "step": 5638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.089156974626617e-07,
+      "loss": 0.7188,
+      "step": 5639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.076093351135001e-07,
+      "loss": 0.7402,
+      "step": 5640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0630569768546946e-07,
+      "loss": 0.8142,
+      "step": 5641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.050047855450711e-07,
+      "loss": 0.7214,
+      "step": 5642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0370659905804786e-07,
+      "loss": 0.9468,
+      "step": 5643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024111385893691e-07,
+      "loss": 0.8545,
+      "step": 5644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0111840450324225e-07,
+      "loss": 0.592,
+      "step": 5645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.998283971631055e-07,
+      "loss": 0.7969,
+      "step": 5646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9854111693163544e-07,
+      "loss": 0.6892,
+      "step": 5647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9725656417073366e-07,
+      "loss": 0.6477,
+      "step": 5648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.959747392415424e-07,
+      "loss": 0.7305,
+      "step": 5649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.946956425044334e-07,
+      "loss": 0.7563,
+      "step": 5650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.934192743190134e-07,
+      "loss": 0.8113,
+      "step": 5651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9214563504411986e-07,
+      "loss": 0.6343,
+      "step": 5652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.90874725037823e-07,
+      "loss": 0.5513,
+      "step": 5653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896065446574292e-07,
+      "loss": 0.7502,
+      "step": 5654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.883410942594733e-07,
+      "loss": 0.7354,
+      "step": 5655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8707837419972516e-07,
+      "loss": 0.8342,
+      "step": 5656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8581838483318524e-07,
+      "loss": 0.7393,
+      "step": 5657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.845611265140891e-07,
+      "loss": 0.6633,
+      "step": 5658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8330659959589944e-07,
+      "loss": 0.8027,
+      "step": 5659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8205480443131873e-07,
+      "loss": 0.6663,
+      "step": 5660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8080574137227203e-07,
+      "loss": 0.6396,
+      "step": 5661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.795594107699262e-07,
+      "loss": 0.6985,
+      "step": 5662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7831581297466993e-07,
+      "loss": 0.7,
+      "step": 5663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7707494833613235e-07,
+      "loss": 0.7231,
+      "step": 5664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7583681720317e-07,
+      "loss": 0.533,
+      "step": 5665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.746014199238689e-07,
+      "loss": 0.7659,
+      "step": 5666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733687568455523e-07,
+      "loss": 0.8213,
+      "step": 5667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7213882831476966e-07,
+      "loss": 0.7676,
+      "step": 5668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.709116346773033e-07,
+      "loss": 0.7854,
+      "step": 5669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.696871762781661e-07,
+      "loss": 0.6135,
+      "step": 5670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684654534616049e-07,
+      "loss": 0.7156,
+      "step": 5671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6724646657109164e-07,
+      "loss": 0.7817,
+      "step": 5672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6603021594933663e-07,
+      "loss": 0.8633,
+      "step": 5673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6481670193827413e-07,
+      "loss": 0.7576,
+      "step": 5674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6360592487907234e-07,
+      "loss": 0.792,
+      "step": 5675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6239788511213003e-07,
+      "loss": 0.665,
+      "step": 5676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611925829770745e-07,
+      "loss": 0.6504,
+      "step": 5677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5999001881276686e-07,
+      "loss": 0.7654,
+      "step": 5678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5879019295729447e-07,
+      "loss": 0.7993,
+      "step": 5679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.575931057479786e-07,
+      "loss": 0.7805,
+      "step": 5680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5639875752136554e-07,
+      "loss": 0.7842,
+      "step": 5681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.552071486132368e-07,
+      "loss": 0.8481,
+      "step": 5682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5401827935860103e-07,
+      "loss": 0.7068,
+      "step": 5683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.528321500916986e-07,
+      "loss": 0.7437,
+      "step": 5684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5164876114599503e-07,
+      "loss": 0.7136,
+      "step": 5685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5046811285419305e-07,
+      "loss": 0.7108,
+      "step": 5686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49290205548216e-07,
+      "loss": 0.6536,
+      "step": 5687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4811503955922246e-07,
+      "loss": 0.73,
+      "step": 5688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.469426152176013e-07,
+      "loss": 0.6072,
+      "step": 5689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4577293285296456e-07,
+      "loss": 0.5918,
+      "step": 5690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.446059927941602e-07,
+      "loss": 0.6357,
+      "step": 5691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4344179536925916e-07,
+      "loss": 0.8135,
+      "step": 5692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.422803409055674e-07,
+      "loss": 0.6174,
+      "step": 5693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.411216297296137e-07,
+      "loss": 0.7451,
+      "step": 5694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3996566216715977e-07,
+      "loss": 0.6858,
+      "step": 5695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3881243854319334e-07,
+      "loss": 0.7576,
+      "step": 5696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3766195918193403e-07,
+      "loss": 0.7007,
+      "step": 5697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3651422440682637e-07,
+      "loss": 0.7693,
+      "step": 5698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.353692345405445e-07,
+      "loss": 0.5952,
+      "step": 5699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3422698990499315e-07,
+      "loss": 0.5916,
+      "step": 5700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.33087490821301e-07,
+      "loss": 0.7695,
+      "step": 5701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3195073760982956e-07,
+      "loss": 0.6978,
+      "step": 5702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3081673059016207e-07,
+      "loss": 0.793,
+      "step": 5703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2968547008111796e-07,
+      "loss": 0.679,
+      "step": 5704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2855695640073727e-07,
+      "loss": 0.6648,
+      "step": 5705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2743118986629177e-07,
+      "loss": 0.6541,
+      "step": 5706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2630817079427936e-07,
+      "loss": 0.708,
+      "step": 5707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251878995004264e-07,
+      "loss": 0.5092,
+      "step": 5708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2407037629968431e-07,
+      "loss": 0.6387,
+      "step": 5709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229556015062373e-07,
+      "loss": 0.6609,
+      "step": 5710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.218435754334891e-07,
+      "loss": 0.5417,
+      "step": 5711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2073429839407855e-07,
+      "loss": 0.6694,
+      "step": 5712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196277706998673e-07,
+      "loss": 0.8066,
+      "step": 5713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.7769,
+      "step": 5714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1742296459062562e-07,
+      "loss": 0.6729,
+      "step": 5715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1632468679545494e-07,
+      "loss": 0.6655,
+      "step": 5716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1522915958520296e-07,
+      "loss": 0.6743,
+      "step": 5717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1413638326786445e-07,
+      "loss": 0.7617,
+      "step": 5718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1304635815066587e-07,
+      "loss": 0.8657,
+      "step": 5719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1195908454005543e-07,
+      "loss": 0.7354,
+      "step": 5720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.108745627417097e-07,
+      "loss": 0.9023,
+      "step": 5721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0979279306053035e-07,
+      "loss": 0.7505,
+      "step": 5722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0871377580064744e-07,
+      "loss": 0.8088,
+      "step": 5723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0763751126541721e-07,
+      "loss": 0.7371,
+      "step": 5724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0656399975741982e-07,
+      "loss": 0.6394,
+      "step": 5725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0549324157846273e-07,
+      "loss": 0.804,
+      "step": 5726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.044252370295774e-07,
+      "loss": 0.6865,
+      "step": 5727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0335998641102583e-07,
+      "loss": 0.8042,
+      "step": 5728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.022974900222907e-07,
+      "loss": 0.7229,
+      "step": 5729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.012377481620853e-07,
+      "loss": 0.7964,
+      "step": 5730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0018076112834128e-07,
+      "loss": 0.5974,
+      "step": 5731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.991265292182243e-07,
+      "loss": 0.7681,
+      "step": 5732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9807505272811946e-07,
+      "loss": 0.6954,
+      "step": 5733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9702633195363918e-07,
+      "loss": 0.7451,
+      "step": 5734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95980367189621e-07,
+      "loss": 0.7085,
+      "step": 5735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9493715873012854e-07,
+      "loss": 0.6289,
+      "step": 5736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.938967068684494e-07,
+      "loss": 0.8062,
+      "step": 5737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9285901189709522e-07,
+      "loss": 0.752,
+      "step": 5738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.91824074107807e-07,
+      "loss": 0.6465,
+      "step": 5739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9079189379154317e-07,
+      "loss": 0.7817,
+      "step": 5740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.897624712384949e-07,
+      "loss": 0.6738,
+      "step": 5741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8873580673807178e-07,
+      "loss": 0.6968,
+      "step": 5742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8771190057891186e-07,
+      "loss": 0.7332,
+      "step": 5743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8669075304887596e-07,
+      "loss": 0.6797,
+      "step": 5744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8567236443504999e-07,
+      "loss": 0.9082,
+      "step": 5745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.846567350237427e-07,
+      "loss": 0.7319,
+      "step": 5746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8364386510049015e-07,
+      "loss": 0.7798,
+      "step": 5747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.826337549500501e-07,
+      "loss": 0.813,
+      "step": 5748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8162640485640425e-07,
+      "loss": 0.8345,
+      "step": 5749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8062181510276167e-07,
+      "loss": 0.6931,
+      "step": 5750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7961998597155083e-07,
+      "loss": 0.8872,
+      "step": 5751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7862091774442757e-07,
+      "loss": 0.8306,
+      "step": 5752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7762461070226834e-07,
+      "loss": 0.7322,
+      "step": 5753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.766310651251779e-07,
+      "loss": 0.6436,
+      "step": 5754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7564028129247846e-07,
+      "loss": 0.8427,
+      "step": 5755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7465225948272268e-07,
+      "loss": 0.663,
+      "step": 5756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7366699997368064e-07,
+      "loss": 0.853,
+      "step": 5757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7268450304234962e-07,
+      "loss": 0.7817,
+      "step": 5758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.717047689649487e-07,
+      "loss": 0.8599,
+      "step": 5759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7072779801692086e-07,
+      "loss": 0.7368,
+      "step": 5760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6975359047293306e-07,
+      "loss": 0.8325,
+      "step": 5761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6878214660687175e-07,
+      "loss": 0.7754,
+      "step": 5762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6781346669185184e-07,
+      "loss": 0.7654,
+      "step": 5763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6684755100020435e-07,
+      "loss": 0.8394,
+      "step": 5764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6588439980349092e-07,
+      "loss": 0.8037,
+      "step": 5765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6492401337248943e-07,
+      "loss": 0.7095,
+      "step": 5766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.639663919772061e-07,
+      "loss": 0.781,
+      "step": 5767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6301153588686336e-07,
+      "loss": 0.7803,
+      "step": 5768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6205944536991203e-07,
+      "loss": 0.7473,
+      "step": 5769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111012069402132e-07,
+      "loss": 0.6895,
+      "step": 5770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6016356212608663e-07,
+      "loss": 0.6982,
+      "step": 5771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592197699322229e-07,
+      "loss": 0.7705,
+      "step": 5772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.582787443777667e-07,
+      "loss": 0.698,
+      "step": 5773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.573404857272809e-07,
+      "loss": 0.7388,
+      "step": 5774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5640499424454448e-07,
+      "loss": 0.7661,
+      "step": 5775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5547227019256596e-07,
+      "loss": 0.7478,
+      "step": 5776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5454231383356666e-07,
+      "loss": 0.792,
+      "step": 5777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.536151254289997e-07,
+      "loss": 0.6614,
+      "step": 5778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.52690705239531e-07,
+      "loss": 0.7361,
+      "step": 5779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5176905352505377e-07,
+      "loss": 0.7437,
+      "step": 5780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5085017054468188e-07,
+      "loss": 0.6616,
+      "step": 5781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4993405655674974e-07,
+      "loss": 0.6785,
+      "step": 5782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4902071181881582e-07,
+      "loss": 0.6792,
+      "step": 5783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4811013658765471e-07,
+      "loss": 0.7744,
+      "step": 5784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4720233111926828e-07,
+      "loss": 0.7441,
+      "step": 5785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4629729566887462e-07,
+      "loss": 0.7754,
+      "step": 5786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4539503049091908e-07,
+      "loss": 0.7871,
+      "step": 5787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4449553583906317e-07,
+      "loss": 0.7006,
+      "step": 5788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.435988119661902e-07,
+      "loss": 0.7207,
+      "step": 5789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4270485912440513e-07,
+      "loss": 0.7695,
+      "step": 5790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4181367756503693e-07,
+      "loss": 0.7041,
+      "step": 5791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4092526753862856e-07,
+      "loss": 0.8308,
+      "step": 5792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 0.6836,
+      "step": 5793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3915676308299154e-07,
+      "loss": 0.7898,
+      "step": 5794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3827666915095962e-07,
+      "loss": 0.6748,
+      "step": 5795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3739934774628427e-07,
+      "loss": 0.8213,
+      "step": 5796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365247991156171e-07,
+      "loss": 0.7502,
+      "step": 5797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3565302350482924e-07,
+      "loss": 0.7202,
+      "step": 5798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3478402115901013e-07,
+      "loss": 0.6787,
+      "step": 5799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3391779232247326e-07,
+      "loss": 0.7649,
+      "step": 5800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3305433723874828e-07,
+      "loss": 0.7615,
+      "step": 5801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3219365615058877e-07,
+      "loss": 0.6445,
+      "step": 5802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3133574929996673e-07,
+      "loss": 0.7939,
+      "step": 5803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3048061692807367e-07,
+      "loss": 0.7036,
+      "step": 5804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2962825927532175e-07,
+      "loss": 0.7698,
+      "step": 5805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2877867658134257e-07,
+      "loss": 0.7837,
+      "step": 5806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2793186908498955e-07,
+      "loss": 0.6782,
+      "step": 5807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2708783702433335e-07,
+      "loss": 0.7095,
+      "step": 5808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.262465806366664e-07,
+      "loss": 0.7295,
+      "step": 5809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2540810015849725e-07,
+      "loss": 0.6195,
+      "step": 5810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2457239582555957e-07,
+      "loss": 0.7246,
+      "step": 5811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23739467872801e-07,
+      "loss": 0.7527,
+      "step": 5812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2290931653439307e-07,
+      "loss": 0.6367,
+      "step": 5813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.220819420437236e-07,
+      "loss": 0.7141,
+      "step": 5814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2125734463340088e-07,
+      "loss": 0.6951,
+      "step": 5815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2043552453525177e-07,
+      "loss": 0.7266,
+      "step": 5816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1961648198032473e-07,
+      "loss": 0.6396,
+      "step": 5817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1880021719888446e-07,
+      "loss": 0.5701,
+      "step": 5818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.179867304204163e-07,
+      "loss": 0.8477,
+      "step": 5819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1717602187362398e-07,
+      "loss": 0.7739,
+      "step": 5820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.163680917864296e-07,
+      "loss": 0.7493,
+      "step": 5821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1556294038597593e-07,
+      "loss": 0.8335,
+      "step": 5822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.147605678986219e-07,
+      "loss": 0.6799,
+      "step": 5823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1396097454994925e-07,
+      "loss": 0.5685,
+      "step": 5824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1316416056475377e-07,
+      "loss": 0.7106,
+      "step": 5825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123701261670529e-07,
+      "loss": 0.6973,
+      "step": 5826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.115788715800803e-07,
+      "loss": 0.73,
+      "step": 5827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1079039702629135e-07,
+      "loss": 0.7236,
+      "step": 5828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1000470272735652e-07,
+      "loss": 0.7832,
+      "step": 5829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.092217889041669e-07,
+      "loss": 0.6584,
+      "step": 5830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0844165577683086e-07,
+      "loss": 0.75,
+      "step": 5831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0766430356467628e-07,
+      "loss": 0.6702,
+      "step": 5832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0688973248624613e-07,
+      "loss": 0.6572,
+      "step": 5833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0611794275930398e-07,
+      "loss": 0.7314,
+      "step": 5834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0534893460083296e-07,
+      "loss": 0.9072,
+      "step": 5835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.045827082270301e-07,
+      "loss": 0.8091,
+      "step": 5836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0381926385331309e-07,
+      "loss": 0.8315,
+      "step": 5837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0305860169431581e-07,
+      "loss": 0.6711,
+      "step": 5838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0230072196389384e-07,
+      "loss": 0.7642,
+      "step": 5839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0154562487511455e-07,
+      "loss": 0.7312,
+      "step": 5840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0079331064026921e-07,
+      "loss": 0.7639,
+      "step": 5841
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0004377947086086e-07,
+      "loss": 0.8325,
+      "step": 5842
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.929703157761317e-08,
+      "loss": 0.771,
+      "step": 5843
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.855306717046931e-08,
+      "loss": 0.7988,
+      "step": 5844
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.78118864585853e-08,
+      "loss": 0.8159,
+      "step": 5845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.70734896503378e-08,
+      "loss": 0.5251,
+      "step": 5846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.633787695331965e-08,
+      "loss": 0.7205,
+      "step": 5847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.560504857434205e-08,
+      "loss": 0.6379,
+      "step": 5848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.487500471943135e-08,
+      "loss": 0.5876,
+      "step": 5849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.414774559383332e-08,
+      "loss": 0.8569,
+      "step": 5850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.34232714020078e-08,
+      "loss": 0.6238,
+      "step": 5851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.270158234763627e-08,
+      "loss": 0.873,
+      "step": 5852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.19826786336131e-08,
+      "loss": 0.6846,
+      "step": 5853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.126656046204996e-08,
+      "loss": 0.7361,
+      "step": 5854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.055322803427913e-08,
+      "loss": 0.7527,
+      "step": 5855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.984268155084575e-08,
+      "loss": 0.8245,
+      "step": 5856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.913492121151335e-08,
+      "loss": 0.8564,
+      "step": 5857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.842994721526054e-08,
+      "loss": 0.9351,
+      "step": 5858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.772775976028547e-08,
+      "loss": 0.7725,
+      "step": 5859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.70283590440002e-08,
+      "loss": 0.7207,
+      "step": 5860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.633174526303634e-08,
+      "loss": 0.7639,
+      "step": 5861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.563791861323723e-08,
+      "loss": 0.6765,
+      "step": 5862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.494687928966794e-08,
+      "loss": 0.8792,
+      "step": 5863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.42586274866053e-08,
+      "loss": 0.7273,
+      "step": 5864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.357316339754562e-08,
+      "loss": 0.6855,
+      "step": 5865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.289048721520143e-08,
+      "loss": 0.7393,
+      "step": 5866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.22105991314992e-08,
+      "loss": 0.7197,
+      "step": 5867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.15334993375838e-08,
+      "loss": 0.7693,
+      "step": 5868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.085918802381298e-08,
+      "loss": 0.6821,
+      "step": 5869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.01876653797662e-08,
+      "loss": 0.7368,
+      "step": 5870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.95189315942324e-08,
+      "loss": 0.6702,
+      "step": 5871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.5835,
+      "step": 5872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.818983134995739e-08,
+      "loss": 0.6335,
+      "step": 5873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.752946526487837e-08,
+      "loss": 0.7036,
+      "step": 5874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.687188878564122e-08,
+      "loss": 0.7065,
+      "step": 5875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.621710209711586e-08,
+      "loss": 0.749,
+      "step": 5876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.556510538339056e-08,
+      "loss": 0.7234,
+      "step": 5877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.491589882776761e-08,
+      "loss": 0.5942,
+      "step": 5878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.426948261276324e-08,
+      "loss": 0.679,
+      "step": 5879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.362585692011426e-08,
+      "loss": 0.7161,
+      "step": 5880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.298502193076817e-08,
+      "loss": 0.7532,
+      "step": 5881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.234697782488864e-08,
+      "loss": 0.8237,
+      "step": 5882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.171172478185661e-08,
+      "loss": 0.7964,
+      "step": 5883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.10792629802659e-08,
+      "loss": 0.9028,
+      "step": 5884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.044959259792871e-08,
+      "loss": 0.8213,
+      "step": 5885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.982271381187012e-08,
+      "loss": 0.708,
+      "step": 5886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.919862679833134e-08,
+      "loss": 0.6946,
+      "step": 5887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.85773317327676e-08,
+      "loss": 0.8398,
+      "step": 5888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.795882878985027e-08,
+      "loss": 0.748,
+      "step": 5889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.734311814346472e-08,
+      "loss": 0.8022,
+      "step": 5890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.673019996671359e-08,
+      "loss": 0.7192,
+      "step": 5891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.612007443191237e-08,
+      "loss": 0.7495,
+      "step": 5892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.551274171059052e-08,
+      "loss": 0.6982,
+      "step": 5893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.49082019734959e-08,
+      "loss": 0.7793,
+      "step": 5894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.430645539058589e-08,
+      "loss": 0.7158,
+      "step": 5895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.37075021310396e-08,
+      "loss": 0.6545,
+      "step": 5896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.311134236324346e-08,
+      "loss": 0.7021,
+      "step": 5897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.251797625480449e-08,
+      "loss": 0.5447,
+      "step": 5898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.192740397253926e-08,
+      "loss": 0.6152,
+      "step": 5899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.133962568248386e-08,
+      "loss": 0.6182,
+      "step": 5900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.075464154988497e-08,
+      "loss": 0.7969,
+      "step": 5901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.01724517392055e-08,
+      "loss": 0.7476,
+      "step": 5902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.959305641412228e-08,
+      "loss": 0.7812,
+      "step": 5903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.901645573752723e-08,
+      "loss": 0.7004,
+      "step": 5904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8442649871526255e-08,
+      "loss": 0.7664,
+      "step": 5905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.787163897743697e-08,
+      "loss": 0.6226,
+      "step": 5906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7303423215797624e-08,
+      "loss": 0.7695,
+      "step": 5907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.673800274635266e-08,
+      "loss": 0.7715,
+      "step": 5908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.617537772806603e-08,
+      "loss": 0.7903,
+      "step": 5909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.561554831911453e-08,
+      "loss": 0.8076,
+      "step": 5910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5058514676888944e-08,
+      "loss": 0.8643,
+      "step": 5911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.450427695799177e-08,
+      "loss": 0.7227,
+      "step": 5912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3952835318243914e-08,
+      "loss": 0.6797,
+      "step": 5913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.340418991267804e-08,
+      "loss": 0.7842,
+      "step": 5914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.285834089553854e-08,
+      "loss": 0.8457,
+      "step": 5915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.23152884202871e-08,
+      "loss": 0.7126,
+      "step": 5916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.177503263959604e-08,
+      "loss": 0.6611,
+      "step": 5917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.123757370535498e-08,
+      "loss": 0.8145,
+      "step": 5918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.070291176866415e-08,
+      "loss": 0.7466,
+      "step": 5919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.017104697983777e-08,
+      "loss": 0.7366,
+      "step": 5920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.964197948840621e-08,
+      "loss": 0.7744,
+      "step": 5921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.911570944311051e-08,
+      "loss": 0.6575,
+      "step": 5922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.859223699190674e-08,
+      "loss": 0.4813,
+      "step": 5923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8071562281963856e-08,
+      "loss": 0.636,
+      "step": 5924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7553685459663657e-08,
+      "loss": 0.7747,
+      "step": 5925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.703860667060301e-08,
+      "loss": 0.5481,
+      "step": 5926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652632605959273e-08,
+      "loss": 0.5956,
+      "step": 5927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.601684377065319e-08,
+      "loss": 0.6536,
+      "step": 5928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.551015994702201e-08,
+      "loss": 0.7954,
+      "step": 5929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.500627473114638e-08,
+      "loss": 0.658,
+      "step": 5930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.450518826469075e-08,
+      "loss": 0.6267,
+      "step": 5931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.40069006885302e-08,
+      "loss": 0.7288,
+      "step": 5932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.351141214275268e-08,
+      "loss": 0.7632,
+      "step": 5933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3018722766661194e-08,
+      "loss": 0.7568,
+      "step": 5934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2528832698769395e-08,
+      "loss": 0.6924,
+      "step": 5935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2041742076806e-08,
+      "loss": 0.6587,
+      "step": 5936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1557451037710364e-08,
+      "loss": 0.6338,
+      "step": 5937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.107595971763689e-08,
+      "loss": 0.6824,
+      "step": 5938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.059726825195398e-08,
+      "loss": 0.7258,
+      "step": 5939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.012137677523842e-08,
+      "loss": 0.7463,
+      "step": 5940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.964828542128429e-08,
+      "loss": 0.804,
+      "step": 5941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.917799432309521e-08,
+      "loss": 0.7263,
+      "step": 5942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.871050361288986e-08,
+      "loss": 0.7722,
+      "step": 5943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8245813422098653e-08,
+      "loss": 0.7124,
+      "step": 5944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.778392388136376e-08,
+      "loss": 0.7727,
+      "step": 5945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.732483512054241e-08,
+      "loss": 0.9224,
+      "step": 5946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6868547268702484e-08,
+      "loss": 0.8027,
+      "step": 5947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.641506045412358e-08,
+      "loss": 0.6033,
+      "step": 5948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.596437480430037e-08,
+      "loss": 0.8867,
+      "step": 5949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.55164904459393e-08,
+      "loss": 0.6333,
+      "step": 5950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 0.6742,
+      "step": 5951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4629126106484614e-08,
+      "loss": 0.6909,
+      "step": 5952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4189646374866994e-08,
+      "loss": 0.8335,
+      "step": 5953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.375296843365683e-08,
+      "loss": 0.7827,
+      "step": 5954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.33190924056237e-08,
+      "loss": 0.7407,
+      "step": 5955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.28880184127478e-08,
+      "loss": 0.751,
+      "step": 5956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2459746576219976e-08,
+      "loss": 0.6814,
+      "step": 5957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2034277016446124e-08,
+      "loss": 0.6392,
+      "step": 5958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.161160985304168e-08,
+      "loss": 0.7832,
+      "step": 5959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.119174520483603e-08,
+      "loss": 0.7002,
+      "step": 5960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0774683189870314e-08,
+      "loss": 0.7896,
+      "step": 5961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.036042392539629e-08,
+      "loss": 0.7217,
+      "step": 5962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.994896752788079e-08,
+      "loss": 0.8196,
+      "step": 5963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9540314112999068e-08,
+      "loss": 0.7253,
+      "step": 5964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.913446379564033e-08,
+      "loss": 0.7827,
+      "step": 5965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8731416689906643e-08,
+      "loss": 0.7417,
+      "step": 5966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8331172909108474e-08,
+      "loss": 0.7183,
+      "step": 5967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7933732565773587e-08,
+      "loss": 0.626,
+      "step": 5968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.753909577163483e-08,
+      "loss": 0.6689,
+      "step": 5969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7147262637643447e-08,
+      "loss": 0.8213,
+      "step": 5970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.67582332739591e-08,
+      "loss": 0.6648,
+      "step": 5971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.637200778995208e-08,
+      "loss": 0.814,
+      "step": 5972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5988586294206643e-08,
+      "loss": 0.6792,
+      "step": 5973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5607968894518775e-08,
+      "loss": 0.9067,
+      "step": 5974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5230155697893998e-08,
+      "loss": 0.6741,
+      "step": 5975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4855146810551788e-08,
+      "loss": 0.8481,
+      "step": 5976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4482942337922255e-08,
+      "loss": 0.8356,
+      "step": 5977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.411354238464725e-08,
+      "loss": 0.8467,
+      "step": 5978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.374694705457814e-08,
+      "loss": 0.7129,
+      "step": 5979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3383156450782486e-08,
+      "loss": 0.674,
+      "step": 5980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3022170675534028e-08,
+      "loss": 0.7856,
+      "step": 5981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2663989830321586e-08,
+      "loss": 0.8254,
+      "step": 5982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.23086140158435e-08,
+      "loss": 0.689,
+      "step": 5983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1956043332010957e-08,
+      "loss": 0.9917,
+      "step": 5984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.160627787794578e-08,
+      "loss": 0.8218,
+      "step": 5985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1259317751980422e-08,
+      "loss": 0.6064,
+      "step": 5986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0915163051660193e-08,
+      "loss": 0.7595,
+      "step": 5987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0573813873739912e-08,
+      "loss": 0.8765,
+      "step": 5988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023527031418726e-08,
+      "loss": 0.6753,
+      "step": 5989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989953246818055e-08,
+      "loss": 0.7871,
+      "step": 5990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9566600430108717e-08,
+      "loss": 0.7932,
+      "step": 5991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9236474293572448e-08,
+      "loss": 0.6267,
+      "step": 5992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8909154151384167e-08,
+      "loss": 0.7998,
+      "step": 5993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858464009556582e-08,
+      "loss": 0.7656,
+      "step": 5994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8262932217351094e-08,
+      "loss": 0.6066,
+      "step": 5995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7944030607185415e-08,
+      "loss": 0.7524,
+      "step": 5996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7627935354725957e-08,
+      "loss": 0.5631,
+      "step": 5997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.73146465488383e-08,
+      "loss": 0.6741,
+      "step": 5998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7004164277600878e-08,
+      "loss": 0.6526,
+      "step": 5999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6696488628303864e-08,
+      "loss": 0.7397,
+      "step": 6000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6391619687446957e-08,
+      "loss": 0.5042,
+      "step": 6001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6089557540739376e-08,
+      "loss": 0.6404,
+      "step": 6002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5790302273104295e-08,
+      "loss": 0.8301,
+      "step": 6003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5493853968675533e-08,
+      "loss": 0.6028,
+      "step": 6004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5200212710794195e-08,
+      "loss": 0.7466,
+      "step": 6005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4909378582015356e-08,
+      "loss": 0.7371,
+      "step": 6006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4621351664105831e-08,
+      "loss": 0.866,
+      "step": 6007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4336132038039741e-08,
+      "loss": 0.7974,
+      "step": 6008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.405371978400516e-08,
+      "loss": 0.7302,
+      "step": 6009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3774114981397469e-08,
+      "loss": 0.6146,
+      "step": 6010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3497317708828228e-08,
+      "loss": 0.6663,
+      "step": 6011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.322332804411297e-08,
+      "loss": 0.5464,
+      "step": 6012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.29521460642823e-08,
+      "loss": 0.7693,
+      "step": 6013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2683771845577453e-08,
+      "loss": 0.7598,
+      "step": 6014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2418205463448075e-08,
+      "loss": 0.7798,
+      "step": 6015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2155446992556664e-08,
+      "loss": 0.8862,
+      "step": 6016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1895496506773019e-08,
+      "loss": 0.812,
+      "step": 6017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1638354079182013e-08,
+      "loss": 0.6829,
+      "step": 6018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1384019782075818e-08,
+      "loss": 0.7642,
+      "step": 6019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1132493686957235e-08,
+      "loss": 0.7769,
+      "step": 6020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0883775864541923e-08,
+      "loss": 0.5686,
+      "step": 6021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0637866384752837e-08,
+      "loss": 0.7236,
+      "step": 6022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0394765316726895e-08,
+      "loss": 0.7368,
+      "step": 6023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0154472728808318e-08,
+      "loss": 0.8257,
+      "step": 6024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.916988688553065e-09,
+      "loss": 0.6951,
+      "step": 6025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.682313262728837e-09,
+      "loss": 0.79,
+      "step": 6026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.45044651730953e-09,
+      "loss": 0.6848,
+      "step": 6027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.221388517485218e-09,
+      "loss": 0.7947,
+      "step": 6028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.995139327651058e-09,
+      "loss": 0.8943,
+      "step": 6029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 0.7622,
+      "step": 6030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.551067631599186e-09,
+      "loss": 0.8472,
+      "step": 6031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.333245250226052e-09,
+      "loss": 0.7964,
+      "step": 6032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.118231928537779e-09,
+      "loss": 0.731,
+      "step": 6033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.906027726981568e-09,
+      "loss": 0.5576,
+      "step": 6034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.696632705217478e-09,
+      "loss": 0.7788,
+      "step": 6035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.49004692211508e-09,
+      "loss": 0.7144,
+      "step": 6036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.286270435753473e-09,
+      "loss": 0.6139,
+      "step": 6037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.085303303422386e-09,
+      "loss": 0.7036,
+      "step": 6038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8871455816221784e-09,
+      "loss": 0.7405,
+      "step": 6039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.691797326062732e-09,
+      "loss": 0.7401,
+      "step": 6040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4992585916634485e-09,
+      "loss": 0.8735,
+      "step": 6041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.309529432556583e-09,
+      "loss": 0.8384,
+      "step": 6042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.122609902081689e-09,
+      "loss": 0.7847,
+      "step": 6043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.938500052788953e-09,
+      "loss": 0.7271,
+      "step": 6044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.757199936439195e-09,
+      "loss": 0.8848,
+      "step": 6045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5787096040027526e-09,
+      "loss": 0.821,
+      "step": 6046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.403029105661705e-09,
+      "loss": 0.8159,
+      "step": 6047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.230158490806547e-09,
+      "loss": 0.6066,
+      "step": 6048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0600978080384e-09,
+      "loss": 0.8018,
+      "step": 6049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.892847105166798e-09,
+      "loss": 0.6198,
+      "step": 6050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.728406429214127e-09,
+      "loss": 0.6333,
+      "step": 6051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.566775826410075e-09,
+      "loss": 0.7307,
+      "step": 6052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.407955342196068e-09,
+      "loss": 0.5641,
+      "step": 6053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.251945021223059e-09,
+      "loss": 0.6826,
+      "step": 6054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0987449073526254e-09,
+      "loss": 0.656,
+      "step": 6055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.948355043653651e-09,
+      "loss": 0.7507,
+      "step": 6056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8007754724078695e-09,
+      "loss": 0.677,
+      "step": 6057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.656006235105425e-09,
+      "loss": 0.8115,
+      "step": 6058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.514047372448204e-09,
+      "loss": 0.7471,
+      "step": 6059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3748989243442832e-09,
+      "loss": 0.6658,
+      "step": 6060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.238560929915702e-09,
+      "loss": 0.5828,
+      "step": 6061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.105033427491799e-09,
+      "loss": 0.7451,
+      "step": 6062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9743164546114366e-09,
+      "loss": 0.6128,
+      "step": 6063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.846410048027437e-09,
+      "loss": 0.7522,
+      "step": 6064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7213142436965935e-09,
+      "loss": 0.79,
+      "step": 6065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5990290767885506e-09,
+      "loss": 0.7402,
+      "step": 6066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4795545816846955e-09,
+      "loss": 0.6265,
+      "step": 6067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3628907919726052e-09,
+      "loss": 0.5334,
+      "step": 6068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.249037740450488e-09,
+      "loss": 0.6631,
+      "step": 6069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1379954591282946e-09,
+      "loss": 0.6143,
+      "step": 6070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.029763979224386e-09,
+      "loss": 0.7502,
+      "step": 6071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.924343331166645e-09,
+      "loss": 0.7078,
+      "step": 6072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8217335445935847e-09,
+      "loss": 0.7405,
+      "step": 6073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7219346483521304e-09,
+      "loss": 0.8379,
+      "step": 6074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6249466704998385e-09,
+      "loss": 0.834,
+      "step": 6075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5307696383048965e-09,
+      "loss": 0.6392,
+      "step": 6076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.439403578242793e-09,
+      "loss": 0.7764,
+      "step": 6077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3508485160007579e-09,
+      "loss": 0.8638,
+      "step": 6078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.265104476476653e-09,
+      "loss": 0.7273,
+      "step": 6079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1821714837745302e-09,
+      "loss": 0.7561,
+      "step": 6080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1020495612112936e-09,
+      "loss": 0.7246,
+      "step": 6081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.024738731311148e-09,
+      "loss": 0.8308,
+      "step": 6082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.502390158111497e-10,
+      "loss": 0.6616,
+      "step": 6083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.785504356556562e-10,
+      "loss": 0.7959,
+      "step": 6084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.096730109985462e-10,
+      "loss": 0.8403,
+      "step": 6085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.436067612032194e-10,
+      "loss": 0.6863,
+      "step": 6086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.803517048459274e-10,
+      "loss": 0.7251,
+      "step": 6087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.199078597091124e-10,
+      "loss": 0.8015,
+      "step": 6088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.622752427847378e-10,
+      "loss": 0.8333,
+      "step": 6089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.074538702776188e-10,
+      "loss": 0.6943,
+      "step": 6090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5544375759876135e-10,
+      "loss": 0.6323,
+      "step": 6091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.06244919372023e-10,
+      "loss": 0.6248,
+      "step": 6092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.598573694274521e-10,
+      "loss": 0.7236,
+      "step": 6093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1628112080683837e-10,
+      "loss": 0.5806,
+      "step": 6094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7551618576149297e-10,
+      "loss": 0.7079,
+      "step": 6095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.375625757522482e-10,
+      "loss": 0.6399,
+      "step": 6096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0242030144834724e-10,
+      "loss": 0.6951,
+      "step": 6097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.700893727307751e-10,
+      "loss": 0.7876,
+      "step": 6098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4056979869003785e-10,
+      "loss": 0.7227,
+      "step": 6099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1386158762283217e-10,
+      "loss": 0.8477,
+      "step": 6100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.996474703981684e-11,
+      "loss": 0.7344,
+      "step": 6101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.887928365784113e-11,
+      "loss": 0.678,
+      "step": 6102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.060520340660624e-11,
+      "loss": 0.8633,
+      "step": 6103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.514251142200387e-11,
+      "loss": 0.4802,
+      "step": 6104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.24912120527776e-11,
+      "loss": 0.6738,
+      "step": 6105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2651308853861566e-11,
+      "loss": 0.7324,
+      "step": 6106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6228045941519874e-12,
+      "loss": 0.7749,
+      "step": 6107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4057012476254016e-12,
+      "loss": 0.8862,
+      "step": 6108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.606,
+      "step": 6109
+    },
+    {
+      "epoch": 1.0,
+      "step": 6109,
+      "total_flos": 359795350251520.0,
+      "train_loss": 0.6446147288954537,
+      "train_runtime": 56242.59,
+      "train_samples_per_second": 5.213,
+      "train_steps_per_second": 0.109
+    }
+  ],
+  "max_steps": 6109,
+  "num_train_epochs": 1,
+  "total_flos": 359795350251520.0,
+  "trial_name": null,
+  "trial_params": null
+}