diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,173424 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 28899,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3068050749711652e-08,
+      "loss": 0.8688,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6136101499423304e-08,
+      "loss": 0.8219,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.920415224913495e-08,
+      "loss": 0.7815,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.227220299884661e-08,
+      "loss": 0.8381,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1534025374855826e-07,
+      "loss": 0.7907,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.384083044982699e-07,
+      "loss": 0.7724,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6147635524798155e-07,
+      "loss": 0.8725,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8454440599769322e-07,
+      "loss": 0.8127,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0761245674740486e-07,
+      "loss": 0.7946,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3068050749711653e-07,
+      "loss": 0.7229,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5374855824682817e-07,
+      "loss": 0.8252,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.768166089965398e-07,
+      "loss": 0.8589,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9988465974625145e-07,
+      "loss": 0.8531,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.229527104959631e-07,
+      "loss": 0.8102,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.460207612456748e-07,
+      "loss": 0.8193,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6908881199538643e-07,
+      "loss": 0.7489,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 0.8626,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.152249134948097e-07,
+      "loss": 0.7346,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.382929642445214e-07,
+      "loss": 0.7275,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6136101499423305e-07,
+      "loss": 0.7147,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.844290657439446e-07,
+      "loss": 0.7494,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.074971164936563e-07,
+      "loss": 0.8446,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.30565167243368e-07,
+      "loss": 0.7745,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.536332179930796e-07,
+      "loss": 0.7441,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.767012687427913e-07,
+      "loss": 0.6761,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.997693194925029e-07,
+      "loss": 0.7584,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.228373702422146e-07,
+      "loss": 0.658,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.459054209919262e-07,
+      "loss": 0.7037,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.689734717416378e-07,
+      "loss": 0.6733,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.920415224913496e-07,
+      "loss": 0.649,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.151095732410613e-07,
+      "loss": 0.6853,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.381776239907729e-07,
+      "loss": 0.6824,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.612456747404845e-07,
+      "loss": 0.6884,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 0.6174,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.073817762399077e-07,
+      "loss": 0.7343,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.304498269896194e-07,
+      "loss": 0.6125,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.53517877739331e-07,
+      "loss": 0.636,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.765859284890428e-07,
+      "loss": 0.7292,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.996539792387544e-07,
+      "loss": 0.7117,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.227220299884661e-07,
+      "loss": 0.6915,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.457900807381777e-07,
+      "loss": 0.6864,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.688581314878893e-07,
+      "loss": 0.784,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.919261822376009e-07,
+      "loss": 0.6817,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0149942329873127e-06,
+      "loss": 0.6942,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0380622837370243e-06,
+      "loss": 0.6231,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.061130334486736e-06,
+      "loss": 0.6275,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0841983852364477e-06,
+      "loss": 0.6551,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1072664359861592e-06,
+      "loss": 0.6451,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1303344867358708e-06,
+      "loss": 0.687,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1534025374855826e-06,
+      "loss": 0.6381,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.6807,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1995386389850058e-06,
+      "loss": 0.6483,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2226066897347174e-06,
+      "loss": 0.6279,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2456747404844292e-06,
+      "loss": 0.6537,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2687427912341408e-06,
+      "loss": 0.598,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2918108419838524e-06,
+      "loss": 0.6759,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.314878892733564e-06,
+      "loss": 0.662,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3379469434832756e-06,
+      "loss": 0.638,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3610149942329876e-06,
+      "loss": 0.5657,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3840830449826992e-06,
+      "loss": 0.6495,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4071510957324108e-06,
+      "loss": 0.6119,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4302191464821226e-06,
+      "loss": 0.6188,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4532871972318341e-06,
+      "loss": 0.6838,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4763552479815457e-06,
+      "loss": 0.6311,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4994232987312573e-06,
+      "loss": 0.5636,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.522491349480969e-06,
+      "loss": 0.6264,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5455594002306807e-06,
+      "loss": 0.6133,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 0.5967,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5916955017301039e-06,
+      "loss": 0.6237,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6147635524798155e-06,
+      "loss": 0.6551,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.637831603229527e-06,
+      "loss": 0.737,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6608996539792389e-06,
+      "loss": 0.622,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6839677047289505e-06,
+      "loss": 0.566,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.707035755478662e-06,
+      "loss": 0.5757,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7301038062283736e-06,
+      "loss": 0.5629,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7531718569780857e-06,
+      "loss": 0.6735,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7762399077277972e-06,
+      "loss": 0.623,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7993079584775088e-06,
+      "loss": 0.6794,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8223760092272204e-06,
+      "loss": 0.5914,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8454440599769322e-06,
+      "loss": 0.5783,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8685121107266438e-06,
+      "loss": 0.6682,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8915801614763554e-06,
+      "loss": 0.6268,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.914648212226067e-06,
+      "loss": 0.6025,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9377162629757786e-06,
+      "loss": 0.6046,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.567,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9838523644752018e-06,
+      "loss": 0.627,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0069204152249138e-06,
+      "loss": 0.5775,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0299884659746254e-06,
+      "loss": 0.6096,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.053056516724337e-06,
+      "loss": 0.5746,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0761245674740485e-06,
+      "loss": 0.601,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09919261822376e-06,
+      "loss": 0.6777,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.122260668973472e-06,
+      "loss": 0.6393,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1453287197231837e-06,
+      "loss": 0.5473,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1683967704728953e-06,
+      "loss": 0.6272,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191464821222607e-06,
+      "loss": 0.6028,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2145328719723185e-06,
+      "loss": 0.6246,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.23760092272203e-06,
+      "loss": 0.6378,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2606689734717417e-06,
+      "loss": 0.5515,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2837370242214537e-06,
+      "loss": 0.5969,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3068050749711653e-06,
+      "loss": 0.6357,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.329873125720877e-06,
+      "loss": 0.5853,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.5958,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3760092272203e-06,
+      "loss": 0.712,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3990772779700116e-06,
+      "loss": 0.5889,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4221453287197232e-06,
+      "loss": 0.6186,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.445213379469435e-06,
+      "loss": 0.6084,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4682814302191464e-06,
+      "loss": 0.5967,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4913494809688584e-06,
+      "loss": 0.5672,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.51441753171857e-06,
+      "loss": 0.5874,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5374855824682816e-06,
+      "loss": 0.5947,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.560553633217993e-06,
+      "loss": 0.6034,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5836216839677048e-06,
+      "loss": 0.5883,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6066897347174164e-06,
+      "loss": 0.6082,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.629757785467128e-06,
+      "loss": 0.6095,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6528258362168395e-06,
+      "loss": 0.5884,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.675893886966551e-06,
+      "loss": 0.5904,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6989619377162636e-06,
+      "loss": 0.6341,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.722029988465975e-06,
+      "loss": 0.576,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7450980392156867e-06,
+      "loss": 0.5747,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7681660899653983e-06,
+      "loss": 0.5649,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.79123414071511e-06,
+      "loss": 0.5261,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8143021914648215e-06,
+      "loss": 0.6007,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.837370242214533e-06,
+      "loss": 0.5391,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.860438292964245e-06,
+      "loss": 0.5888,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8835063437139567e-06,
+      "loss": 0.6502,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9065743944636683e-06,
+      "loss": 0.579,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.92964244521338e-06,
+      "loss": 0.564,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9527104959630915e-06,
+      "loss": 0.6042,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.975778546712803e-06,
+      "loss": 0.6019,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9988465974625146e-06,
+      "loss": 0.5592,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0219146482122262e-06,
+      "loss": 0.556,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.044982698961938e-06,
+      "loss": 0.6108,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0680507497116494e-06,
+      "loss": 0.5933,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0911188004613614e-06,
+      "loss": 0.5521,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.114186851211073e-06,
+      "loss": 0.6413,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.5869,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.160322952710496e-06,
+      "loss": 0.5761,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1833910034602078e-06,
+      "loss": 0.5589,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2064590542099194e-06,
+      "loss": 0.5545,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.229527104959631e-06,
+      "loss": 0.5561,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2525951557093425e-06,
+      "loss": 0.5835,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.275663206459054e-06,
+      "loss": 0.6095,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.298731257208766e-06,
+      "loss": 0.571,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3217993079584777e-06,
+      "loss": 0.6157,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3448673587081893e-06,
+      "loss": 0.5737,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.367935409457901e-06,
+      "loss": 0.5768,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3910034602076125e-06,
+      "loss": 0.6176,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414071510957324e-06,
+      "loss": 0.6372,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4371395617070357e-06,
+      "loss": 0.4918,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4602076124567473e-06,
+      "loss": 0.5809,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4832756632064597e-06,
+      "loss": 0.6979,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5063437139561713e-06,
+      "loss": 0.6494,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.5995,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5524798154555945e-06,
+      "loss": 0.5768,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.575547866205306e-06,
+      "loss": 0.5295,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5986159169550177e-06,
+      "loss": 0.5785,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6216839677047292e-06,
+      "loss": 0.5285,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.644752018454441e-06,
+      "loss": 0.5814,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667820069204153e-06,
+      "loss": 0.5469,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6908881199538644e-06,
+      "loss": 0.5501,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.713956170703576e-06,
+      "loss": 0.6395,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7370242214532876e-06,
+      "loss": 0.5967,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.760092272202999e-06,
+      "loss": 0.5942,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.783160322952711e-06,
+      "loss": 0.5657,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8062283737024224e-06,
+      "loss": 0.6082,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.829296424452134e-06,
+      "loss": 0.6603,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.852364475201846e-06,
+      "loss": 0.5902,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.875432525951557e-06,
+      "loss": 0.5545,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.898500576701269e-06,
+      "loss": 0.5973,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.6232,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.944636678200692e-06,
+      "loss": 0.5621,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9677047289504035e-06,
+      "loss": 0.6079,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9907727797001155e-06,
+      "loss": 0.6027,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0138408304498275e-06,
+      "loss": 0.6303,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.036908881199539e-06,
+      "loss": 0.6178,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.059976931949251e-06,
+      "loss": 0.4888,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.083044982698962e-06,
+      "loss": 0.5574,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.106113033448674e-06,
+      "loss": 0.5306,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.129181084198385e-06,
+      "loss": 0.5829,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.152249134948097e-06,
+      "loss": 0.6121,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.175317185697808e-06,
+      "loss": 0.6066,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.19838523644752e-06,
+      "loss": 0.6099,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.221453287197232e-06,
+      "loss": 0.6106,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.244521337946944e-06,
+      "loss": 0.5339,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2675893886966554e-06,
+      "loss": 0.5403,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2906574394463675e-06,
+      "loss": 0.5903,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.6212,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.336793540945791e-06,
+      "loss": 0.5947,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.359861591695503e-06,
+      "loss": 0.6407,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.382929642445214e-06,
+      "loss": 0.5595,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.405997693194926e-06,
+      "loss": 0.7067,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.429065743944637e-06,
+      "loss": 0.5452,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.452133794694349e-06,
+      "loss": 0.5266,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.47520184544406e-06,
+      "loss": 0.6571,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.498269896193772e-06,
+      "loss": 0.5971,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.521337946943483e-06,
+      "loss": 0.5634,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.544405997693195e-06,
+      "loss": 0.61,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.567474048442907e-06,
+      "loss": 0.5849,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5905420991926185e-06,
+      "loss": 0.6606,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6136101499423305e-06,
+      "loss": 0.6295,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636678200692042e-06,
+      "loss": 0.594,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659746251441754e-06,
+      "loss": 0.5552,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.682814302191465e-06,
+      "loss": 0.5873,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.6457,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.728950403690888e-06,
+      "loss": 0.6297,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7520184544406e-06,
+      "loss": 0.5873,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.775086505190312e-06,
+      "loss": 0.6035,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.798154555940023e-06,
+      "loss": 0.6257,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.821222606689735e-06,
+      "loss": 0.5451,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8442906574394464e-06,
+      "loss": 0.6223,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8673587081891585e-06,
+      "loss": 0.6082,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89042675893887e-06,
+      "loss": 0.5964,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913494809688582e-06,
+      "loss": 0.5796,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.936562860438293e-06,
+      "loss": 0.5859,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959630911188005e-06,
+      "loss": 0.5654,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982698961937717e-06,
+      "loss": 0.5383,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.005767012687428e-06,
+      "loss": 0.5488,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.02883506343714e-06,
+      "loss": 0.5676,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.051903114186851e-06,
+      "loss": 0.6033,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.074971164936563e-06,
+      "loss": 0.6576,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.6083,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.121107266435986e-06,
+      "loss": 0.539,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1441753171856975e-06,
+      "loss": 0.5625,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1672433679354095e-06,
+      "loss": 0.6185,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.190311418685121e-06,
+      "loss": 0.6286,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.213379469434833e-06,
+      "loss": 0.5204,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.236447520184545e-06,
+      "loss": 0.5764,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.259515570934256e-06,
+      "loss": 0.6224,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.282583621683968e-06,
+      "loss": 0.6024,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.305651672433679e-06,
+      "loss": 0.5369,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.328719723183391e-06,
+      "loss": 0.5696,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.351787773933102e-06,
+      "loss": 0.6006,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.374855824682814e-06,
+      "loss": 0.5041,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.397923875432527e-06,
+      "loss": 0.6122,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.420991926182238e-06,
+      "loss": 0.5798,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.44405997693195e-06,
+      "loss": 0.6518,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4671280276816615e-06,
+      "loss": 0.6011,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 0.5759,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5132641291810855e-06,
+      "loss": 0.5236,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.536332179930797e-06,
+      "loss": 0.6063,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.559400230680509e-06,
+      "loss": 0.5463,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.58246828143022e-06,
+      "loss": 0.5262,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.605536332179932e-06,
+      "loss": 0.5687,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.628604382929643e-06,
+      "loss": 0.5944,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.651672433679355e-06,
+      "loss": 0.5795,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.674740484429066e-06,
+      "loss": 0.5192,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.697808535178778e-06,
+      "loss": 0.6532,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72087658592849e-06,
+      "loss": 0.6318,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.743944636678201e-06,
+      "loss": 0.5994,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.767012687427913e-06,
+      "loss": 0.5624,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7900807381776246e-06,
+      "loss": 0.5678,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8131487889273366e-06,
+      "loss": 0.5007,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.836216839677048e-06,
+      "loss": 0.565,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.85928489042676e-06,
+      "loss": 0.6007,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.5855,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.905420991926183e-06,
+      "loss": 0.6182,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.928489042675895e-06,
+      "loss": 0.629,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.951557093425606e-06,
+      "loss": 0.5304,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.974625144175318e-06,
+      "loss": 0.626,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.997693194925029e-06,
+      "loss": 0.6371,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.020761245674741e-06,
+      "loss": 0.6232,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0438292964244525e-06,
+      "loss": 0.5565,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0668973471741645e-06,
+      "loss": 0.5423,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.089965397923876e-06,
+      "loss": 0.5709,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.113033448673588e-06,
+      "loss": 0.6264,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.136101499423299e-06,
+      "loss": 0.5018,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.159169550173011e-06,
+      "loss": 0.6363,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.182237600922723e-06,
+      "loss": 0.6869,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.205305651672434e-06,
+      "loss": 0.5834,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.228373702422146e-06,
+      "loss": 0.6307,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.251441753171857e-06,
+      "loss": 0.6646,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.6036,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.29757785467128e-06,
+      "loss": 0.5939,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.320645905420992e-06,
+      "loss": 0.5598,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3437139561707036e-06,
+      "loss": 0.601,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3667820069204156e-06,
+      "loss": 0.5607,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.389850057670128e-06,
+      "loss": 0.6138,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.412918108419839e-06,
+      "loss": 0.6068,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.435986159169551e-06,
+      "loss": 0.5635,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.459054209919262e-06,
+      "loss": 0.5549,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.482122260668974e-06,
+      "loss": 0.5945,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.505190311418685e-06,
+      "loss": 0.5876,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.528258362168397e-06,
+      "loss": 0.6259,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.551326412918108e-06,
+      "loss": 0.603,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.57439446366782e-06,
+      "loss": 0.5906,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.597462514417532e-06,
+      "loss": 0.6004,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6205305651672435e-06,
+      "loss": 0.5882,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6435986159169555e-06,
+      "loss": 0.6169,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.618,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.689734717416379e-06,
+      "loss": 0.6075,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.71280276816609e-06,
+      "loss": 0.6255,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.735870818915802e-06,
+      "loss": 0.5803,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.758938869665513e-06,
+      "loss": 0.5513,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.782006920415225e-06,
+      "loss": 0.6414,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.805074971164937e-06,
+      "loss": 0.624,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.828143021914648e-06,
+      "loss": 0.5614,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.85121107266436e-06,
+      "loss": 0.5843,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.874279123414071e-06,
+      "loss": 0.6046,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.897347174163783e-06,
+      "loss": 0.5971,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9204152249134946e-06,
+      "loss": 0.5833,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.943483275663207e-06,
+      "loss": 0.6025,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9665513264129194e-06,
+      "loss": 0.5649,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.989619377162631e-06,
+      "loss": 0.6536,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.012687427912343e-06,
+      "loss": 0.6254,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.035755478662054e-06,
+      "loss": 0.5917,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.6098,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.081891580161478e-06,
+      "loss": 0.5938,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.104959630911189e-06,
+      "loss": 0.5564,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.128027681660901e-06,
+      "loss": 0.5607,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.151095732410612e-06,
+      "loss": 0.6505,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.174163783160324e-06,
+      "loss": 0.6461,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.197231833910035e-06,
+      "loss": 0.6176,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.220299884659747e-06,
+      "loss": 0.6637,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2433679354094585e-06,
+      "loss": 0.542,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2664359861591705e-06,
+      "loss": 0.5408,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.289504036908882e-06,
+      "loss": 0.51,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.312572087658594e-06,
+      "loss": 0.5863,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.335640138408306e-06,
+      "loss": 0.5574,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.358708189158017e-06,
+      "loss": 0.548,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.381776239907729e-06,
+      "loss": 0.6278,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.40484429065744e-06,
+      "loss": 0.6009,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.427912341407152e-06,
+      "loss": 0.6701,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.549,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.474048442906575e-06,
+      "loss": 0.5523,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.497116493656286e-06,
+      "loss": 0.5933,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.520184544405998e-06,
+      "loss": 0.6212,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5432525951557104e-06,
+      "loss": 0.6067,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.566320645905422e-06,
+      "loss": 0.5824,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.589388696655134e-06,
+      "loss": 0.6088,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.612456747404845e-06,
+      "loss": 0.6132,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.635524798154556e-06,
+      "loss": 0.5754,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.658592848904268e-06,
+      "loss": 0.6044,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.68166089965398e-06,
+      "loss": 0.6255,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.704728950403692e-06,
+      "loss": 0.7428,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.727797001153404e-06,
+      "loss": 0.5693,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.750865051903114e-06,
+      "loss": 0.5808,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.773933102652826e-06,
+      "loss": 0.614,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.797001153402538e-06,
+      "loss": 0.5602,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.82006920415225e-06,
+      "loss": 0.6444,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.6268,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.866205305651673e-06,
+      "loss": 0.614,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.889273356401385e-06,
+      "loss": 0.5249,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.912341407151097e-06,
+      "loss": 0.6189,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.935409457900807e-06,
+      "loss": 0.5009,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.958477508650519e-06,
+      "loss": 0.6201,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.981545559400231e-06,
+      "loss": 0.5543,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.004613610149943e-06,
+      "loss": 0.5443,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.027681660899655e-06,
+      "loss": 0.5941,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.050749711649365e-06,
+      "loss": 0.5515,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.073817762399077e-06,
+      "loss": 0.5867,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.09688581314879e-06,
+      "loss": 0.5641,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.119953863898501e-06,
+      "loss": 0.6327,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.143021914648212e-06,
+      "loss": 0.6151,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.166089965397924e-06,
+      "loss": 0.5735,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.189158016147636e-06,
+      "loss": 0.6033,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.212226066897348e-06,
+      "loss": 0.6162,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.562,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.25836216839677e-06,
+      "loss": 0.5291,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.281430219146482e-06,
+      "loss": 0.5697,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.304498269896194e-06,
+      "loss": 0.5524,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.327566320645906e-06,
+      "loss": 0.6041,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.350634371395616e-06,
+      "loss": 0.5694,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.373702422145328e-06,
+      "loss": 0.5936,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.39677047289504e-06,
+      "loss": 0.6335,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.419838523644753e-06,
+      "loss": 0.6206,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.442906574394465e-06,
+      "loss": 0.6197,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.465974625144177e-06,
+      "loss": 0.595,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.489042675893889e-06,
+      "loss": 0.6505,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5121107266436e-06,
+      "loss": 0.6079,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.535178777393311e-06,
+      "loss": 0.5822,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.558246828143023e-06,
+      "loss": 0.6124,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.581314878892735e-06,
+      "loss": 0.6082,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.604382929642447e-06,
+      "loss": 0.6272,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.6291,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.65051903114187e-06,
+      "loss": 0.5661,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.673587081891581e-06,
+      "loss": 0.6096,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.696655132641293e-06,
+      "loss": 0.6148,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.719723183391005e-06,
+      "loss": 0.6231,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.742791234140716e-06,
+      "loss": 0.5733,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.765859284890428e-06,
+      "loss": 0.6084,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.78892733564014e-06,
+      "loss": 0.5853,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.811995386389852e-06,
+      "loss": 0.6062,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.835063437139562e-06,
+      "loss": 0.5459,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.858131487889274e-06,
+      "loss": 0.5339,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.881199538638986e-06,
+      "loss": 0.6479,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.904267589388698e-06,
+      "loss": 0.5922,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92733564013841e-06,
+      "loss": 0.5644,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.95040369088812e-06,
+      "loss": 0.5467,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.973471741637832e-06,
+      "loss": 0.6827,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.996539792387544e-06,
+      "loss": 0.5977,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 0.6676,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.042675893886967e-06,
+      "loss": 0.5817,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.065743944636679e-06,
+      "loss": 0.5815,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.08881199538639e-06,
+      "loss": 0.5881,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.111880046136103e-06,
+      "loss": 0.5689,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.134948096885815e-06,
+      "loss": 0.6062,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.158016147635525e-06,
+      "loss": 0.6029,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.181084198385237e-06,
+      "loss": 0.6095,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.204152249134949e-06,
+      "loss": 0.5957,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.227220299884661e-06,
+      "loss": 0.5352,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.250288350634371e-06,
+      "loss": 0.5674,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.273356401384083e-06,
+      "loss": 0.5674,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.296424452133795e-06,
+      "loss": 0.5669,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.319492502883507e-06,
+      "loss": 0.5671,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.34256055363322e-06,
+      "loss": 0.6338,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.36562860438293e-06,
+      "loss": 0.5447,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.388696655132642e-06,
+      "loss": 0.615,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.5939,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.434832756632066e-06,
+      "loss": 0.6015,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.457900807381776e-06,
+      "loss": 0.593,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.480968858131488e-06,
+      "loss": 0.5283,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5040369088812e-06,
+      "loss": 0.5847,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.527104959630912e-06,
+      "loss": 0.6218,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.550173010380624e-06,
+      "loss": 0.6455,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.573241061130335e-06,
+      "loss": 0.679,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.596309111880047e-06,
+      "loss": 0.6762,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.619377162629759e-06,
+      "loss": 0.4986,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.64244521337947e-06,
+      "loss": 0.6281,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.665513264129181e-06,
+      "loss": 0.5906,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.688581314878893e-06,
+      "loss": 0.5186,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.711649365628605e-06,
+      "loss": 0.5764,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.734717416378317e-06,
+      "loss": 0.5989,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.757785467128029e-06,
+      "loss": 0.6163,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.78085351787774e-06,
+      "loss": 0.5994,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.5988,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.826989619377163e-06,
+      "loss": 0.6394,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.850057670126875e-06,
+      "loss": 0.5669,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.873125720876586e-06,
+      "loss": 0.6046,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.896193771626298e-06,
+      "loss": 0.5566,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.91926182237601e-06,
+      "loss": 0.5993,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.942329873125722e-06,
+      "loss": 0.6174,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965397923875434e-06,
+      "loss": 0.6251,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988465974625144e-06,
+      "loss": 0.6034,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0011534025374856e-05,
+      "loss": 0.5873,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0034602076124568e-05,
+      "loss": 0.5573,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.005767012687428e-05,
+      "loss": 0.6151,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0080738177623992e-05,
+      "loss": 0.5874,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0103806228373702e-05,
+      "loss": 0.5994,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0126874279123416e-05,
+      "loss": 0.624,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0149942329873126e-05,
+      "loss": 0.5765,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0173010380622838e-05,
+      "loss": 0.6084,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.6281,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0219146482122262e-05,
+      "loss": 0.5691,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0242214532871973e-05,
+      "loss": 0.6072,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0265282583621685e-05,
+      "loss": 0.6119,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0288350634371395e-05,
+      "loss": 0.5892,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0311418685121109e-05,
+      "loss": 0.6803,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0334486735870819e-05,
+      "loss": 0.5607,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0357554786620531e-05,
+      "loss": 0.6123,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0380622837370241e-05,
+      "loss": 0.6354,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0403690888119955e-05,
+      "loss": 0.6622,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0426758938869665e-05,
+      "loss": 0.5845,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0449826989619377e-05,
+      "loss": 0.5677,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.047289504036909e-05,
+      "loss": 0.6343,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0495963091118801e-05,
+      "loss": 0.5461,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0519031141868512e-05,
+      "loss": 0.6185,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0542099192618225e-05,
+      "loss": 0.5614,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0565167243367936e-05,
+      "loss": 0.6118,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.5775,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0611303344867358e-05,
+      "loss": 0.5971,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0634371395617072e-05,
+      "loss": 0.588,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0657439446366782e-05,
+      "loss": 0.6951,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0680507497116494e-05,
+      "loss": 0.5948,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0703575547866205e-05,
+      "loss": 0.6289,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0726643598615918e-05,
+      "loss": 0.6712,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0749711649365629e-05,
+      "loss": 0.6094,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.077277970011534e-05,
+      "loss": 0.5908,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0795847750865054e-05,
+      "loss": 0.5875,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0818915801614765e-05,
+      "loss": 0.6668,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0841983852364477e-05,
+      "loss": 0.4543,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0865051903114187e-05,
+      "loss": 0.6243,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.08881199538639e-05,
+      "loss": 0.6158,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0911188004613611e-05,
+      "loss": 0.5856,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0934256055363323e-05,
+      "loss": 0.5982,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0957324106113035e-05,
+      "loss": 0.5578,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0980392156862747e-05,
+      "loss": 0.6159,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1003460207612457e-05,
+      "loss": 0.5853,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1026528258362171e-05,
+      "loss": 0.6851,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1049596309111881e-05,
+      "loss": 0.634,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1072664359861593e-05,
+      "loss": 0.5813,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1095732410611304e-05,
+      "loss": 0.5726,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1118800461361017e-05,
+      "loss": 0.6211,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1141868512110728e-05,
+      "loss": 0.6194,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.116493656286044e-05,
+      "loss": 0.552,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.118800461361015e-05,
+      "loss": 0.6245,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1211072664359864e-05,
+      "loss": 0.6015,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1234140715109574e-05,
+      "loss": 0.6546,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1257208765859286e-05,
+      "loss": 0.6252,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1280276816608996e-05,
+      "loss": 0.5955,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.130334486735871e-05,
+      "loss": 0.5841,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.132641291810842e-05,
+      "loss": 0.5723,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1349480968858132e-05,
+      "loss": 0.6166,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1372549019607844e-05,
+      "loss": 0.6033,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1395617070357556e-05,
+      "loss": 0.6394,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1418685121107267e-05,
+      "loss": 0.6397,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.144175317185698e-05,
+      "loss": 0.6296,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.146482122260669e-05,
+      "loss": 0.6124,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1487889273356403e-05,
+      "loss": 0.6099,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1510957324106113e-05,
+      "loss": 0.5645,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1534025374855827e-05,
+      "loss": 0.6235,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1557093425605537e-05,
+      "loss": 0.5821,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1580161476355249e-05,
+      "loss": 0.5996,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.160322952710496e-05,
+      "loss": 0.6314,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1626297577854673e-05,
+      "loss": 0.5777,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1649365628604383e-05,
+      "loss": 0.6065,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1672433679354095e-05,
+      "loss": 0.6487,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1695501730103806e-05,
+      "loss": 0.6185,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.171856978085352e-05,
+      "loss": 0.5873,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.174163783160323e-05,
+      "loss": 0.6044,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.6647,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1787773933102654e-05,
+      "loss": 0.5956,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1810841983852366e-05,
+      "loss": 0.5893,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1833910034602076e-05,
+      "loss": 0.5776,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.185697808535179e-05,
+      "loss": 0.6577,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.18800461361015e-05,
+      "loss": 0.6182,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1903114186851212e-05,
+      "loss": 0.5747,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1926182237600923e-05,
+      "loss": 0.5737,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1949250288350636e-05,
+      "loss": 0.5168,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1972318339100347e-05,
+      "loss": 0.5558,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1995386389850059e-05,
+      "loss": 0.6167,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2018454440599769e-05,
+      "loss": 0.5905,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2041522491349483e-05,
+      "loss": 0.6752,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2064590542099193e-05,
+      "loss": 0.6092,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2087658592848905e-05,
+      "loss": 0.5013,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2110726643598615e-05,
+      "loss": 0.5839,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2133794694348329e-05,
+      "loss": 0.6904,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 0.5897,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2179930795847751e-05,
+      "loss": 0.6361,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2202998846597463e-05,
+      "loss": 0.6888,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2226066897347175e-05,
+      "loss": 0.5741,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2249134948096886e-05,
+      "loss": 0.586,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2272202998846598e-05,
+      "loss": 0.6197,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.229527104959631e-05,
+      "loss": 0.6225,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2318339100346022e-05,
+      "loss": 0.5966,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2341407151095734e-05,
+      "loss": 0.6712,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2364475201845446e-05,
+      "loss": 0.6227,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2387543252595158e-05,
+      "loss": 0.6231,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2410611303344868e-05,
+      "loss": 0.4725,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2433679354094582e-05,
+      "loss": 0.5848,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2456747404844292e-05,
+      "loss": 0.6184,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2479815455594004e-05,
+      "loss": 0.6117,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2502883506343714e-05,
+      "loss": 0.5894,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2525951557093428e-05,
+      "loss": 0.639,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 0.5987,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.257208765859285e-05,
+      "loss": 0.5592,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.259515570934256e-05,
+      "loss": 0.5746,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2618223760092274e-05,
+      "loss": 0.5331,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2641291810841985e-05,
+      "loss": 0.5384,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2664359861591697e-05,
+      "loss": 0.5933,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2687427912341407e-05,
+      "loss": 0.6098,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.271049596309112e-05,
+      "loss": 0.5883,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2733564013840831e-05,
+      "loss": 0.6294,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2756632064590543e-05,
+      "loss": 0.5983,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2779700115340255e-05,
+      "loss": 0.601,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2802768166089967e-05,
+      "loss": 0.6289,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2825836216839677e-05,
+      "loss": 0.6337,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2848904267589391e-05,
+      "loss": 0.6371,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2871972318339102e-05,
+      "loss": 0.5963,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2895040369088814e-05,
+      "loss": 0.5671,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2918108419838524e-05,
+      "loss": 0.6073,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.6148,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2964244521337948e-05,
+      "loss": 0.6092,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.298731257208766e-05,
+      "loss": 0.6155,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.301038062283737e-05,
+      "loss": 0.5778,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3033448673587084e-05,
+      "loss": 0.6164,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3056516724336794e-05,
+      "loss": 0.5435,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3079584775086506e-05,
+      "loss": 0.5786,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3102652825836217e-05,
+      "loss": 0.7074,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.312572087658593e-05,
+      "loss": 0.6386,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.314878892733564e-05,
+      "loss": 0.6,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3171856978085353e-05,
+      "loss": 0.6549,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3194925028835065e-05,
+      "loss": 0.5706,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3217993079584777e-05,
+      "loss": 0.5793,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3241061130334487e-05,
+      "loss": 0.6675,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.32641291810842e-05,
+      "loss": 0.609,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3287197231833911e-05,
+      "loss": 0.6229,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3310265282583623e-05,
+      "loss": 0.5573,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6197,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3356401384083047e-05,
+      "loss": 0.6128,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3379469434832757e-05,
+      "loss": 0.6068,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.340253748558247e-05,
+      "loss": 0.6347,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.342560553633218e-05,
+      "loss": 0.6241,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3448673587081893e-05,
+      "loss": 0.5982,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3471741637831604e-05,
+      "loss": 0.6474,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3494809688581316e-05,
+      "loss": 0.5938,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3517877739331026e-05,
+      "loss": 0.6056,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.354094579008074e-05,
+      "loss": 0.5895,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.356401384083045e-05,
+      "loss": 0.5993,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3587081891580162e-05,
+      "loss": 0.5448,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3610149942329874e-05,
+      "loss": 0.6429,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3633217993079586e-05,
+      "loss": 0.6645,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3656286043829296e-05,
+      "loss": 0.5733,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.367935409457901e-05,
+      "loss": 0.5863,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.370242214532872e-05,
+      "loss": 0.6154,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 0.5838,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3748558246828143e-05,
+      "loss": 0.5758,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3771626297577856e-05,
+      "loss": 0.5267,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3794694348327567e-05,
+      "loss": 0.6884,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3817762399077279e-05,
+      "loss": 0.5913,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3840830449826989e-05,
+      "loss": 0.5432,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3863898500576703e-05,
+      "loss": 0.611,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3886966551326415e-05,
+      "loss": 0.6041,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3910034602076125e-05,
+      "loss": 0.6088,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3933102652825839e-05,
+      "loss": 0.569,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.395617070357555e-05,
+      "loss": 0.6112,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3979238754325261e-05,
+      "loss": 0.6335,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4002306805074971e-05,
+      "loss": 0.5702,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4025374855824685e-05,
+      "loss": 0.5963,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4048442906574396e-05,
+      "loss": 0.5653,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4071510957324108e-05,
+      "loss": 0.6306,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.409457900807382e-05,
+      "loss": 0.5545,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.5924,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4140715109573242e-05,
+      "loss": 0.5921,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4163783160322956e-05,
+      "loss": 0.6634,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4186851211072666e-05,
+      "loss": 0.6369,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4209919261822378e-05,
+      "loss": 0.5821,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4232987312572088e-05,
+      "loss": 0.6971,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4256055363321802e-05,
+      "loss": 0.6645,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4279123414071512e-05,
+      "loss": 0.6256,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4302191464821224e-05,
+      "loss": 0.6512,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4325259515570935e-05,
+      "loss": 0.5854,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4348327566320648e-05,
+      "loss": 0.6454,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4371395617070359e-05,
+      "loss": 0.5885,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.439446366782007e-05,
+      "loss": 0.589,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4417531718569781e-05,
+      "loss": 0.6447,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4440599769319495e-05,
+      "loss": 0.6421,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4463667820069205e-05,
+      "loss": 0.5968,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4486735870818917e-05,
+      "loss": 0.5734,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 0.5917,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4532871972318341e-05,
+      "loss": 0.6018,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4555940023068051e-05,
+      "loss": 0.6125,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4579008073817763e-05,
+      "loss": 0.6293,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4602076124567475e-05,
+      "loss": 0.5982,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4625144175317187e-05,
+      "loss": 0.6685,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4648212226066898e-05,
+      "loss": 0.6101,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4671280276816611e-05,
+      "loss": 0.6588,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4694348327566322e-05,
+      "loss": 0.5576,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4717416378316034e-05,
+      "loss": 0.6664,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4740484429065744e-05,
+      "loss": 0.6408,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4763552479815458e-05,
+      "loss": 0.6508,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4786620530565168e-05,
+      "loss": 0.5998,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.480968858131488e-05,
+      "loss": 0.6382,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.483275663206459e-05,
+      "loss": 0.6463,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4855824682814304e-05,
+      "loss": 0.6309,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4878892733564014e-05,
+      "loss": 0.6204,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 0.6178,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4925028835063437e-05,
+      "loss": 0.5899,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.494809688581315e-05,
+      "loss": 0.6093,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.497116493656286e-05,
+      "loss": 0.6029,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4994232987312573e-05,
+      "loss": 0.6271,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5017301038062285e-05,
+      "loss": 0.6232,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5040369088811997e-05,
+      "loss": 0.6801,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5063437139561707e-05,
+      "loss": 0.6314,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5086505190311421e-05,
+      "loss": 0.6517,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5109573241061131e-05,
+      "loss": 0.5637,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5132641291810843e-05,
+      "loss": 0.6428,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5155709342560554e-05,
+      "loss": 0.562,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5178777393310267e-05,
+      "loss": 0.6562,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5201845444059978e-05,
+      "loss": 0.5892,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.522491349480969e-05,
+      "loss": 0.6341,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.52479815455594e-05,
+      "loss": 0.5492,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5271049596309112e-05,
+      "loss": 0.6526,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.6337,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5317185697808536e-05,
+      "loss": 0.6077,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5340253748558246e-05,
+      "loss": 0.6249,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.536332179930796e-05,
+      "loss": 0.5794,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5386389850057674e-05,
+      "loss": 0.6146,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5409457900807384e-05,
+      "loss": 0.6542,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5432525951557094e-05,
+      "loss": 0.6226,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5455594002306808e-05,
+      "loss": 0.6916,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5478662053056518e-05,
+      "loss": 0.5854,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.550173010380623e-05,
+      "loss": 0.6542,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5524798154555942e-05,
+      "loss": 0.6422,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5547866205305653e-05,
+      "loss": 0.5676,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5570934256055366e-05,
+      "loss": 0.6272,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5594002306805077e-05,
+      "loss": 0.6605,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.561707035755479e-05,
+      "loss": 0.6423,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.56401384083045e-05,
+      "loss": 0.6278,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.566320645905421e-05,
+      "loss": 0.6493,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 0.6446,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5709342560553635e-05,
+      "loss": 0.6165,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5732410611303345e-05,
+      "loss": 0.6358,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.575547866205306e-05,
+      "loss": 0.5982,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.577854671280277e-05,
+      "loss": 0.6686,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5801614763552483e-05,
+      "loss": 0.6056,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5824682814302193e-05,
+      "loss": 0.6223,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5847750865051904e-05,
+      "loss": 0.6047,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5870818915801614e-05,
+      "loss": 0.604,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5893886966551328e-05,
+      "loss": 0.6066,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5916955017301038e-05,
+      "loss": 0.5805,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5940023068050752e-05,
+      "loss": 0.6219,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5963091118800462e-05,
+      "loss": 0.6049,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5986159169550176e-05,
+      "loss": 0.583,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6009227220299886e-05,
+      "loss": 0.637,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.60322952710496e-05,
+      "loss": 0.5895,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.605536332179931e-05,
+      "loss": 0.6031,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 0.6041,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.610149942329873e-05,
+      "loss": 0.6411,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6124567474048444e-05,
+      "loss": 0.5428,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6147635524798155e-05,
+      "loss": 0.6805,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.617070357554787e-05,
+      "loss": 0.6193,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.619377162629758e-05,
+      "loss": 0.6109,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6216839677047293e-05,
+      "loss": 0.6369,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6239907727797003e-05,
+      "loss": 0.6469,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6262975778546713e-05,
+      "loss": 0.5904,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6286043829296423e-05,
+      "loss": 0.6321,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6309111880046137e-05,
+      "loss": 0.5967,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6332179930795848e-05,
+      "loss": 0.6186,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.635524798154556e-05,
+      "loss": 0.6364,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.637831603229527e-05,
+      "loss": 0.6065,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6401384083044985e-05,
+      "loss": 0.5878,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6424452133794696e-05,
+      "loss": 0.6278,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.644752018454441e-05,
+      "loss": 0.6647,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.6067,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.649365628604383e-05,
+      "loss": 0.6133,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.651672433679354e-05,
+      "loss": 0.555,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6539792387543254e-05,
+      "loss": 0.5838,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6562860438292964e-05,
+      "loss": 0.5664,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6585928489042678e-05,
+      "loss": 0.6218,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6608996539792388e-05,
+      "loss": 0.6046,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6632064590542102e-05,
+      "loss": 0.6287,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6655132641291812e-05,
+      "loss": 0.6086,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6678200692041523e-05,
+      "loss": 0.6328,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6701268742791233e-05,
+      "loss": 0.6164,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6724336793540947e-05,
+      "loss": 0.567,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6747404844290657e-05,
+      "loss": 0.6542,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.677047289504037e-05,
+      "loss": 0.6484,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.679354094579008e-05,
+      "loss": 0.625,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6816608996539795e-05,
+      "loss": 0.6358,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6839677047289505e-05,
+      "loss": 0.6048,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 0.5625,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.688581314878893e-05,
+      "loss": 0.6619,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.690888119953864e-05,
+      "loss": 0.6362,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6931949250288353e-05,
+      "loss": 0.5484,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6955017301038063e-05,
+      "loss": 0.6497,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6978085351787777e-05,
+      "loss": 0.5922,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7001153402537487e-05,
+      "loss": 0.6416,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.70242214532872e-05,
+      "loss": 0.6034,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.704728950403691e-05,
+      "loss": 0.645,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7070357554786622e-05,
+      "loss": 0.6205,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7093425605536332e-05,
+      "loss": 0.5853,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7116493656286046e-05,
+      "loss": 0.5928,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7139561707035756e-05,
+      "loss": 0.6047,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.716262975778547e-05,
+      "loss": 0.6179,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.718569780853518e-05,
+      "loss": 0.65,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7208765859284894e-05,
+      "loss": 0.593,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7231833910034604e-05,
+      "loss": 0.6254,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 0.5838,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7277970011534028e-05,
+      "loss": 0.6168,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.730103806228374e-05,
+      "loss": 0.6453,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.732410611303345e-05,
+      "loss": 0.6091,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7347174163783163e-05,
+      "loss": 0.549,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7370242214532873e-05,
+      "loss": 0.6128,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7393310265282587e-05,
+      "loss": 0.6091,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7416378316032297e-05,
+      "loss": 0.6341,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.743944636678201e-05,
+      "loss": 0.5739,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.746251441753172e-05,
+      "loss": 0.6417,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.748558246828143e-05,
+      "loss": 0.6202,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.750865051903114e-05,
+      "loss": 0.6437,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7531718569780855e-05,
+      "loss": 0.6124,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7554786620530566e-05,
+      "loss": 0.6536,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.757785467128028e-05,
+      "loss": 0.6114,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.760092272202999e-05,
+      "loss": 0.5823,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7623990772779703e-05,
+      "loss": 0.7207,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.6693,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7670126874279124e-05,
+      "loss": 0.6728,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7693194925028838e-05,
+      "loss": 0.7416,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7716262975778548e-05,
+      "loss": 0.7101,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7739331026528258e-05,
+      "loss": 0.6234,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7762399077277972e-05,
+      "loss": 0.5579,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7785467128027682e-05,
+      "loss": 0.6094,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7808535178777396e-05,
+      "loss": 0.655,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7831603229527106e-05,
+      "loss": 0.582,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785467128027682e-05,
+      "loss": 0.5864,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.787773933102653e-05,
+      "loss": 0.5437,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.790080738177624e-05,
+      "loss": 0.5813,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.792387543252595e-05,
+      "loss": 0.6286,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7946943483275665e-05,
+      "loss": 0.6958,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7970011534025375e-05,
+      "loss": 0.6427,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.799307958477509e-05,
+      "loss": 0.6463,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.80161476355248e-05,
+      "loss": 0.6435,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8039215686274513e-05,
+      "loss": 0.6673,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8062283737024223e-05,
+      "loss": 0.6131,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8085351787773933e-05,
+      "loss": 0.6728,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8108419838523644e-05,
+      "loss": 0.6397,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8131487889273357e-05,
+      "loss": 0.6308,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8154555940023068e-05,
+      "loss": 0.6616,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.817762399077278e-05,
+      "loss": 0.6397,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8200692041522492e-05,
+      "loss": 0.572,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8223760092272205e-05,
+      "loss": 0.6219,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8246828143021916e-05,
+      "loss": 0.5301,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.826989619377163e-05,
+      "loss": 0.5681,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.829296424452134e-05,
+      "loss": 0.6464,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.831603229527105e-05,
+      "loss": 0.6616,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.833910034602076e-05,
+      "loss": 0.6343,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8362168396770474e-05,
+      "loss": 0.6211,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8385236447520184e-05,
+      "loss": 0.6482,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8408304498269898e-05,
+      "loss": 0.6218,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 0.5935,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8454440599769322e-05,
+      "loss": 0.5843,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8477508650519033e-05,
+      "loss": 0.6052,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8500576701268743e-05,
+      "loss": 0.6352,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8523644752018457e-05,
+      "loss": 0.4999,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8546712802768167e-05,
+      "loss": 0.6219,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.856978085351788e-05,
+      "loss": 0.5925,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.859284890426759e-05,
+      "loss": 0.6009,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8615916955017305e-05,
+      "loss": 0.6447,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8638985005767015e-05,
+      "loss": 0.6054,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8662053056516725e-05,
+      "loss": 0.6439,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.868512110726644e-05,
+      "loss": 0.6017,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.870818915801615e-05,
+      "loss": 0.6192,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.873125720876586e-05,
+      "loss": 0.6208,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8754325259515573e-05,
+      "loss": 0.5902,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8777393310265284e-05,
+      "loss": 0.6334,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8800461361014997e-05,
+      "loss": 0.6681,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.5834,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.884659746251442e-05,
+      "loss": 0.6058,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.886966551326413e-05,
+      "loss": 0.6307,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8892733564013842e-05,
+      "loss": 0.6342,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8915801614763552e-05,
+      "loss": 0.6756,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8938869665513266e-05,
+      "loss": 0.6172,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8961937716262976e-05,
+      "loss": 0.667,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.898500576701269e-05,
+      "loss": 0.6749,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.90080738177624e-05,
+      "loss": 0.6377,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9031141868512114e-05,
+      "loss": 0.5809,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9054209919261824e-05,
+      "loss": 0.6601,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9077277970011535e-05,
+      "loss": 0.607,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.910034602076125e-05,
+      "loss": 0.6201,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.912341407151096e-05,
+      "loss": 0.5965,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.914648212226067e-05,
+      "loss": 0.6368,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9169550173010383e-05,
+      "loss": 0.7114,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9192618223760093e-05,
+      "loss": 0.6777,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 0.5906,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9238754325259517e-05,
+      "loss": 0.5915,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.926182237600923e-05,
+      "loss": 0.5814,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928489042675894e-05,
+      "loss": 0.5763,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.930795847750865e-05,
+      "loss": 0.5998,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9331026528258362e-05,
+      "loss": 0.592,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9354094579008075e-05,
+      "loss": 0.6177,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9377162629757786e-05,
+      "loss": 0.5915,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.94002306805075e-05,
+      "loss": 0.633,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.942329873125721e-05,
+      "loss": 0.6154,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9446366782006923e-05,
+      "loss": 0.6384,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9469434832756634e-05,
+      "loss": 0.6065,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9492502883506344e-05,
+      "loss": 0.6368,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9515570934256058e-05,
+      "loss": 0.6196,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9538638985005768e-05,
+      "loss": 0.6694,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956170703575548e-05,
+      "loss": 0.5917,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9584775086505192e-05,
+      "loss": 0.6664,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.6043,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630911188004616e-05,
+      "loss": 0.6531,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653979238754327e-05,
+      "loss": 0.6631,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967704728950404e-05,
+      "loss": 0.6602,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970011534025375e-05,
+      "loss": 0.5757,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.972318339100346e-05,
+      "loss": 0.6141,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.974625144175317e-05,
+      "loss": 0.6106,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9769319492502885e-05,
+      "loss": 0.6329,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792387543252595e-05,
+      "loss": 0.5719,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981545559400231e-05,
+      "loss": 0.6984,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983852364475202e-05,
+      "loss": 0.5697,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9861591695501733e-05,
+      "loss": 0.6181,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9884659746251443e-05,
+      "loss": 0.6491,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9907727797001154e-05,
+      "loss": 0.5997,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9930795847750867e-05,
+      "loss": 0.6249,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9953863898500578e-05,
+      "loss": 0.6091,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9976931949250288e-05,
+      "loss": 0.6131,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.7025,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999937199722e-05,
+      "loss": 0.5629,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999748798883e-05,
+      "loss": 0.5857,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999434797484e-05,
+      "loss": 0.5861,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998995195536e-05,
+      "loss": 0.672,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998429993043e-05,
+      "loss": 0.648,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997739190003e-05,
+      "loss": 0.5541,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999692278643e-05,
+      "loss": 0.5923,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995980782344e-05,
+      "loss": 0.6595,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994913177745e-05,
+      "loss": 0.62,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999371997265e-05,
+      "loss": 0.6185,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992401167072e-05,
+      "loss": 0.6381,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999990956761032e-05,
+      "loss": 0.6134,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998938675454e-05,
+      "loss": 0.6173,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998769114763e-05,
+      "loss": 0.6066,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999985869940307e-05,
+      "loss": 0.6712,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999983923132605e-05,
+      "loss": 0.6483,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999981850724543e-05,
+      "loss": 0.6342,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997965271615e-05,
+      "loss": 0.6097,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997732910745e-05,
+      "loss": 0.5917,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999974879898477e-05,
+      "loss": 0.5911,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999972305089265e-05,
+      "loss": 0.6604,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999969604679833e-05,
+      "loss": 0.6487,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999966778670224e-05,
+      "loss": 0.649,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999963827060473e-05,
+      "loss": 0.6463,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999960749850616e-05,
+      "loss": 0.62,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999957547040695e-05,
+      "loss": 0.6184,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999954218630743e-05,
+      "loss": 0.6783,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999950764620807e-05,
+      "loss": 0.6992,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999947185010932e-05,
+      "loss": 0.6557,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999943479801155e-05,
+      "loss": 0.5475,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993964899153e-05,
+      "loss": 0.6,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999935692582107e-05,
+      "loss": 0.6011,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993161057293e-05,
+      "loss": 0.6273,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992740296405e-05,
+      "loss": 0.6245,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992306975552e-05,
+      "loss": 0.6089,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999918610947396e-05,
+      "loss": 0.6294,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991402653974e-05,
+      "loss": 0.5993,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999093165326e-05,
+      "loss": 0.5611,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999990448092604e-05,
+      "loss": 0.5928,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989951972012e-05,
+      "loss": 0.6241,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99998944329149e-05,
+      "loss": 0.6923,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999889220510447e-05,
+      "loss": 0.6057,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999883882506824e-05,
+      "loss": 0.6728,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99998784189041e-05,
+      "loss": 0.5682,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999872829702346e-05,
+      "loss": 0.6469,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999867114901626e-05,
+      "loss": 0.6001,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999861274502014e-05,
+      "loss": 0.6517,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999855308503587e-05,
+      "loss": 0.6722,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999849216906417e-05,
+      "loss": 0.6759,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999842999710578e-05,
+      "loss": 0.6355,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999836656916155e-05,
+      "loss": 0.6111,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999983018852322e-05,
+      "loss": 0.5623,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982359453186e-05,
+      "loss": 0.6585,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999816874942156e-05,
+      "loss": 0.5937,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999981002975419e-05,
+      "loss": 0.5569,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999980305896805e-05,
+      "loss": 0.5927,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999795962583824e-05,
+      "loss": 0.6029,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999788740601602e-05,
+      "loss": 0.5627,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999781393021474e-05,
+      "loss": 0.5585,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999773919843527e-05,
+      "loss": 0.7275,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999766321067864e-05,
+      "loss": 0.6455,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999758596694575e-05,
+      "loss": 0.6341,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999750746723756e-05,
+      "loss": 0.6742,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999974277115551e-05,
+      "loss": 0.6555,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999734669989933e-05,
+      "loss": 0.602,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999972644322713e-05,
+      "loss": 0.6961,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99997180908672e-05,
+      "loss": 0.6027,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999709612910254e-05,
+      "loss": 0.7094,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999701009356394e-05,
+      "loss": 0.6362,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999692280205728e-05,
+      "loss": 0.5936,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999683425458372e-05,
+      "loss": 0.5753,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999674445114428e-05,
+      "loss": 0.7095,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999665339174013e-05,
+      "loss": 0.5795,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999965610763724e-05,
+      "loss": 0.5989,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999964675050423e-05,
+      "loss": 0.6711,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999637267775092e-05,
+      "loss": 0.6151,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999627659449954e-05,
+      "loss": 0.6377,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999961792552893e-05,
+      "loss": 0.4995,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999608066012147e-05,
+      "loss": 0.6632,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999598080899726e-05,
+      "loss": 0.6837,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999958797019179e-05,
+      "loss": 0.621,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999577733888476e-05,
+      "loss": 0.7024,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99995673719899e-05,
+      "loss": 0.5918,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99995568844962e-05,
+      "loss": 0.6246,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999546271407503e-05,
+      "loss": 0.6001,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999535532723947e-05,
+      "loss": 0.5925,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999524668445662e-05,
+      "loss": 0.5719,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951367857279e-05,
+      "loss": 0.615,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999502563105466e-05,
+      "loss": 0.6223,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999949132204383e-05,
+      "loss": 0.6582,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999947995538802e-05,
+      "loss": 0.6462,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999468463138184e-05,
+      "loss": 0.586,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999456845294467e-05,
+      "loss": 0.5691,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999445101857006e-05,
+      "loss": 0.5803,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999943323282596e-05,
+      "loss": 0.6807,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999421238201465e-05,
+      "loss": 0.5987,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999409117983685e-05,
+      "loss": 0.6545,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999396872172766e-05,
+      "loss": 0.6079,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999384500768863e-05,
+      "loss": 0.5841,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999937200377213e-05,
+      "loss": 0.6691,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999359381182724e-05,
+      "loss": 0.6173,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999346633000806e-05,
+      "loss": 0.5135,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999933375922653e-05,
+      "loss": 0.5944,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999320759860065e-05,
+      "loss": 0.5901,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999930763490157e-05,
+      "loss": 0.6283,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999294384351212e-05,
+      "loss": 0.5302,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999281008209153e-05,
+      "loss": 0.639,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999926750647557e-05,
+      "loss": 0.6373,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999253879150626e-05,
+      "loss": 0.6273,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999240126234492e-05,
+      "loss": 0.6085,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999226247727343e-05,
+      "loss": 0.6422,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999212243629352e-05,
+      "loss": 0.6263,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999919811394069e-05,
+      "loss": 0.6109,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999183858661548e-05,
+      "loss": 0.5901,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999169477792094e-05,
+      "loss": 0.6066,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999915497133251e-05,
+      "loss": 0.6027,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999914033928298e-05,
+      "loss": 0.6831,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999125581643686e-05,
+      "loss": 0.5994,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999110698414817e-05,
+      "loss": 0.6065,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999095689596556e-05,
+      "loss": 0.6054,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999080555189092e-05,
+      "loss": 0.623,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999906529519262e-05,
+      "loss": 0.5922,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999049909607327e-05,
+      "loss": 0.6223,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999034398433405e-05,
+      "loss": 0.5919,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999018761671053e-05,
+      "loss": 0.6236,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999002999320464e-05,
+      "loss": 0.6417,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998987111381835e-05,
+      "loss": 0.6726,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998971097855372e-05,
+      "loss": 0.6671,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999895495874127e-05,
+      "loss": 0.6838,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998938694039732e-05,
+      "loss": 0.6473,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998922303750968e-05,
+      "loss": 0.5939,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998905787875176e-05,
+      "loss": 0.6231,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999888914641257e-05,
+      "loss": 0.608,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998872379363356e-05,
+      "loss": 0.6365,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998855486727743e-05,
+      "loss": 0.6425,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998838468505947e-05,
+      "loss": 0.6333,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998821324698178e-05,
+      "loss": 0.6577,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998804055304658e-05,
+      "loss": 0.6299,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998786660325592e-05,
+      "loss": 0.5737,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998769139761213e-05,
+      "loss": 0.5919,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998751493611728e-05,
+      "loss": 0.6412,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998733721877367e-05,
+      "loss": 0.6433,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999871582455835e-05,
+      "loss": 0.5894,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99986978016549e-05,
+      "loss": 0.6847,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999867965316725e-05,
+      "loss": 0.625,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998661379095622e-05,
+      "loss": 0.5765,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998642979440243e-05,
+      "loss": 0.6402,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999862445420135e-05,
+      "loss": 0.6086,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998605803379175e-05,
+      "loss": 0.5768,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998587026973954e-05,
+      "loss": 0.6126,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998568124985916e-05,
+      "loss": 0.6327,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998549097415303e-05,
+      "loss": 0.6224,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998529944262356e-05,
+      "loss": 0.5881,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999851066552731e-05,
+      "loss": 0.608,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999849126121041e-05,
+      "loss": 0.6372,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99984717313119e-05,
+      "loss": 0.6207,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998452075832028e-05,
+      "loss": 0.6452,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998432294771036e-05,
+      "loss": 0.5519,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998412388129173e-05,
+      "loss": 0.6234,
+      "step": 1026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999839235590669e-05,
+      "loss": 0.6547,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999837219810384e-05,
+      "loss": 0.6413,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998351914720876e-05,
+      "loss": 0.6053,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998331505758056e-05,
+      "loss": 0.7326,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998310971215623e-05,
+      "loss": 0.6475,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999829031109385e-05,
+      "loss": 0.6166,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998269525392992e-05,
+      "loss": 0.5656,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998248614113306e-05,
+      "loss": 0.6445,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999822757725506e-05,
+      "loss": 0.5941,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998206414818513e-05,
+      "loss": 0.6269,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998185126803936e-05,
+      "loss": 0.6461,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999816371321159e-05,
+      "loss": 0.5848,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814217404175e-05,
+      "loss": 0.577,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998120509294686e-05,
+      "loss": 0.6272,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998098718970668e-05,
+      "loss": 0.6585,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998076803069967e-05,
+      "loss": 0.5984,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999805476159286e-05,
+      "loss": 0.6527,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998032594539634e-05,
+      "loss": 0.6174,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998010301910553e-05,
+      "loss": 0.6314,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99979878837059e-05,
+      "loss": 0.6062,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997965339925965e-05,
+      "loss": 0.5906,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999794267057102e-05,
+      "loss": 0.599,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997919875641362e-05,
+      "loss": 0.6074,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997896955137265e-05,
+      "loss": 0.5769,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997873909059022e-05,
+      "loss": 0.6082,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997850737406925e-05,
+      "loss": 0.5591,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999782744018126e-05,
+      "loss": 0.5781,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997804017382327e-05,
+      "loss": 0.6643,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997780469010412e-05,
+      "loss": 0.622,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997756795065816e-05,
+      "loss": 0.6038,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997732995548832e-05,
+      "loss": 0.6385,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997709070459766e-05,
+      "loss": 0.6285,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997685019798913e-05,
+      "loss": 0.5804,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999766084356657e-05,
+      "loss": 0.6658,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997636541763054e-05,
+      "loss": 0.6438,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999761211438866e-05,
+      "loss": 0.7395,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997587561443698e-05,
+      "loss": 0.5896,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999756288292848e-05,
+      "loss": 0.6534,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997538078843308e-05,
+      "loss": 0.5993,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99975131491885e-05,
+      "loss": 0.6632,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997488093964367e-05,
+      "loss": 0.5934,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999746291317122e-05,
+      "loss": 0.6539,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997437606809386e-05,
+      "loss": 0.6263,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999741217487917e-05,
+      "loss": 0.6272,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99973866173809e-05,
+      "loss": 0.5742,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999736093431489e-05,
+      "loss": 0.6632,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997335125681473e-05,
+      "loss": 0.6167,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999730919148096e-05,
+      "loss": 0.5898,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997283131713687e-05,
+      "loss": 0.5846,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999725694637998e-05,
+      "loss": 0.5808,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999723063548016e-05,
+      "loss": 0.6302,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997204199014572e-05,
+      "loss": 0.5761,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999717763698353e-05,
+      "loss": 0.637,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997150949387384e-05,
+      "loss": 0.6496,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999712413622646e-05,
+      "loss": 0.5936,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997097197501097e-05,
+      "loss": 0.6147,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997070133211635e-05,
+      "loss": 0.6695,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997042943358408e-05,
+      "loss": 0.5961,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997015627941763e-05,
+      "loss": 0.6793,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996988186962044e-05,
+      "loss": 0.6399,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999696062041959e-05,
+      "loss": 0.6205,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996932928314753e-05,
+      "loss": 0.6,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996905110647882e-05,
+      "loss": 0.5199,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996877167419317e-05,
+      "loss": 0.576,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999684909862942e-05,
+      "loss": 0.5918,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996820904278533e-05,
+      "loss": 0.6429,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999679258436702e-05,
+      "loss": 0.6272,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999676413889523e-05,
+      "loss": 0.637,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996735567863522e-05,
+      "loss": 0.638,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996706871272256e-05,
+      "loss": 0.6298,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996678049121793e-05,
+      "loss": 0.6125,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996649101412492e-05,
+      "loss": 0.6232,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996620028144722e-05,
+      "loss": 0.6177,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999659082931884e-05,
+      "loss": 0.5976,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996561504935223e-05,
+      "loss": 0.5838,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999653205499423e-05,
+      "loss": 0.6295,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999650247949623e-05,
+      "loss": 0.6121,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996472778441606e-05,
+      "loss": 0.7076,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996442951830722e-05,
+      "loss": 0.6179,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996412999663954e-05,
+      "loss": 0.5868,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999638292194168e-05,
+      "loss": 0.6063,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996352718664273e-05,
+      "loss": 0.6571,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996322389832122e-05,
+      "loss": 0.6197,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996291935445598e-05,
+      "loss": 0.5196,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996261355505088e-05,
+      "loss": 0.6865,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996230650010975e-05,
+      "loss": 0.5929,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996199818963645e-05,
+      "loss": 0.5866,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996168862363487e-05,
+      "loss": 0.64,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999613778021089e-05,
+      "loss": 0.6556,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996106572506238e-05,
+      "loss": 0.6819,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999607523924993e-05,
+      "loss": 0.6903,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996043780442355e-05,
+      "loss": 0.6198,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996012196083915e-05,
+      "loss": 0.5866,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995980486175e-05,
+      "loss": 0.5967,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999594865071601e-05,
+      "loss": 0.6387,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999591668970734e-05,
+      "loss": 0.6618,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995884603149403e-05,
+      "loss": 0.6002,
+      "step": 1123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995852391042594e-05,
+      "loss": 0.6496,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999582005338732e-05,
+      "loss": 0.6508,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995787590183982e-05,
+      "loss": 0.6126,
+      "step": 1126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995755001432996e-05,
+      "loss": 0.6394,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995722287134766e-05,
+      "loss": 0.6472,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995689447289702e-05,
+      "loss": 0.6195,
+      "step": 1129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995656481898225e-05,
+      "loss": 0.6172,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999562339096074e-05,
+      "loss": 0.6159,
+      "step": 1131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995590174477665e-05,
+      "loss": 0.5817,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995556832449417e-05,
+      "loss": 0.6237,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995523364876415e-05,
+      "loss": 0.6509,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999548977175908e-05,
+      "loss": 0.6076,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995456053097834e-05,
+      "loss": 0.5968,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99954222088931e-05,
+      "loss": 0.6076,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995388239145305e-05,
+      "loss": 0.5881,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995354143854867e-05,
+      "loss": 0.6338,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995319923022228e-05,
+      "loss": 0.6033,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999528557664781e-05,
+      "loss": 0.682,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995251104732047e-05,
+      "loss": 0.5722,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995216507275366e-05,
+      "loss": 0.6182,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995181784278207e-05,
+      "loss": 0.6185,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995146935741007e-05,
+      "loss": 0.6354,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99951119616642e-05,
+      "loss": 0.6122,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995076862048228e-05,
+      "loss": 0.6514,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995041636893532e-05,
+      "loss": 0.5691,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999500628620055e-05,
+      "loss": 0.6312,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994970809969732e-05,
+      "loss": 0.616,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994935208201518e-05,
+      "loss": 0.5865,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999489948089636e-05,
+      "loss": 0.5609,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994863628054706e-05,
+      "loss": 0.6126,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994827649677006e-05,
+      "loss": 0.6278,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994791545763707e-05,
+      "loss": 0.6056,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994755316315266e-05,
+      "loss": 0.6167,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994718961332142e-05,
+      "loss": 0.639,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994682480814784e-05,
+      "loss": 0.5881,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994645874763657e-05,
+      "loss": 0.6065,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999460914317922e-05,
+      "loss": 0.6014,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999457228606193e-05,
+      "loss": 0.607,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994535303412252e-05,
+      "loss": 0.6031,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999449819523065e-05,
+      "loss": 0.6952,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994460961517595e-05,
+      "loss": 0.6407,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994423602273544e-05,
+      "loss": 0.7117,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999438611749898e-05,
+      "loss": 0.6517,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994348507194364e-05,
+      "loss": 0.5892,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999431077136017e-05,
+      "loss": 0.6371,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999427290999687e-05,
+      "loss": 0.6233,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994234923104948e-05,
+      "loss": 0.6679,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999419681068487e-05,
+      "loss": 0.6226,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994158572737127e-05,
+      "loss": 0.5715,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999412020926219e-05,
+      "loss": 0.5995,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994081720260542e-05,
+      "loss": 0.6066,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999404310573267e-05,
+      "loss": 0.6401,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994004365679054e-05,
+      "loss": 0.6484,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993965500100185e-05,
+      "loss": 0.6014,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993926508996548e-05,
+      "loss": 0.6439,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993887392368636e-05,
+      "loss": 0.5691,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999384815021694e-05,
+      "loss": 0.666,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999380878254195e-05,
+      "loss": 0.6355,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999376928934416e-05,
+      "loss": 0.6765,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993729670624073e-05,
+      "loss": 0.5967,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993689926382175e-05,
+      "loss": 0.6401,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993650056618977e-05,
+      "loss": 0.6482,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993610061334972e-05,
+      "loss": 0.6167,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999356994053067e-05,
+      "loss": 0.6641,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993529694206564e-05,
+      "loss": 0.5992,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993489322363166e-05,
+      "loss": 0.6429,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993448825000982e-05,
+      "loss": 0.6409,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993408202120522e-05,
+      "loss": 0.6353,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99933674537223e-05,
+      "loss": 0.6506,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993326579806817e-05,
+      "loss": 0.5934,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999328558037459e-05,
+      "loss": 0.6282,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993244455426147e-05,
+      "loss": 0.6348,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993203204961985e-05,
+      "loss": 0.5893,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993161828982635e-05,
+      "loss": 0.6757,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999312032748861e-05,
+      "loss": 0.6796,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993078700480438e-05,
+      "loss": 0.639,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993036947958636e-05,
+      "loss": 0.5941,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992995069923728e-05,
+      "loss": 0.6285,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992953066376244e-05,
+      "loss": 0.6429,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999291093731671e-05,
+      "loss": 0.5547,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992868682745654e-05,
+      "loss": 0.6385,
+      "step": 1204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999282630266361e-05,
+      "loss": 0.5699,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992783797071105e-05,
+      "loss": 0.6034,
+      "step": 1206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992741165968677e-05,
+      "loss": 0.626,
+      "step": 1207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999269840935686e-05,
+      "loss": 0.6401,
+      "step": 1208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992655527236193e-05,
+      "loss": 0.6453,
+      "step": 1209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999261251960721e-05,
+      "loss": 0.5851,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999256938647046e-05,
+      "loss": 0.616,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992526127826474e-05,
+      "loss": 0.6311,
+      "step": 1212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99924827436758e-05,
+      "loss": 0.6094,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992439234018983e-05,
+      "loss": 0.6125,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999239559885657e-05,
+      "loss": 0.6254,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999235183818911e-05,
+      "loss": 0.6467,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992307952017152e-05,
+      "loss": 0.658,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992263940341243e-05,
+      "loss": 0.6182,
+      "step": 1218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999221980316194e-05,
+      "loss": 0.6451,
+      "step": 1219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99921755404798e-05,
+      "loss": 0.6567,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992131152295375e-05,
+      "loss": 0.5652,
+      "step": 1221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999208663860922e-05,
+      "loss": 0.5986,
+      "step": 1222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99920419994219e-05,
+      "loss": 0.5654,
+      "step": 1223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999199723473397e-05,
+      "loss": 0.6702,
+      "step": 1224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991952344545996e-05,
+      "loss": 0.6023,
+      "step": 1225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991907328858543e-05,
+      "loss": 0.65,
+      "step": 1226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991862187672174e-05,
+      "loss": 0.6332,
+      "step": 1227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991816920987456e-05,
+      "loss": 0.6814,
+      "step": 1228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991771528804954e-05,
+      "loss": 0.5838,
+      "step": 1229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999172601112525e-05,
+      "loss": 0.6273,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99916803679489e-05,
+      "loss": 0.6181,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991634599276488e-05,
+      "loss": 0.6081,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991588705108586e-05,
+      "loss": 0.6544,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991542685445774e-05,
+      "loss": 0.6299,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999149654028862e-05,
+      "loss": 0.5473,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991450269637712e-05,
+      "loss": 0.6382,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991403873493628e-05,
+      "loss": 0.5725,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991357351856956e-05,
+      "loss": 0.6037,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999131070472827e-05,
+      "loss": 0.6063,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991263932108166e-05,
+      "loss": 0.6265,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999121703399723e-05,
+      "loss": 0.6229,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991170010396043e-05,
+      "loss": 0.6666,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99911228613052e-05,
+      "loss": 0.5153,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99910755867253e-05,
+      "loss": 0.6094,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999102818665693e-05,
+      "loss": 0.6178,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999098066110068e-05,
+      "loss": 0.598,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990933010057163e-05,
+      "loss": 0.608,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999088523352696e-05,
+      "loss": 0.607,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990837331510684e-05,
+      "loss": 0.5945,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999078930400893e-05,
+      "loss": 0.6976,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990741151022302e-05,
+      "loss": 0.6506,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990692872551407e-05,
+      "loss": 0.6786,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990644468596846e-05,
+      "loss": 0.6516,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990595939159237e-05,
+      "loss": 0.5804,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990547284239178e-05,
+      "loss": 0.6427,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990498503837287e-05,
+      "loss": 0.6214,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999044959795418e-05,
+      "loss": 0.5656,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999040056659046e-05,
+      "loss": 0.6168,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990351409746754e-05,
+      "loss": 0.5473,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990302127423672e-05,
+      "loss": 0.5793,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990252719621837e-05,
+      "loss": 0.6036,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990203186341867e-05,
+      "loss": 0.6011,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990153527584387e-05,
+      "loss": 0.5673,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990103743350018e-05,
+      "loss": 0.5726,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990053833639386e-05,
+      "loss": 0.6287,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999000379845312e-05,
+      "loss": 0.6171,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989953637791847e-05,
+      "loss": 0.6305,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989903351656195e-05,
+      "loss": 0.7131,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99898529400468e-05,
+      "loss": 0.5711,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998980240296429e-05,
+      "loss": 0.6381,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989751740409306e-05,
+      "loss": 0.6188,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989700952382476e-05,
+      "loss": 0.6614,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989650038884446e-05,
+      "loss": 0.6246,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989598999915845e-05,
+      "loss": 0.633,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989547835477328e-05,
+      "loss": 0.5732,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998949654556953e-05,
+      "loss": 0.6506,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989445130193096e-05,
+      "loss": 0.596,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989393589348672e-05,
+      "loss": 0.6014,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989341923036903e-05,
+      "loss": 0.6011,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989290131258443e-05,
+      "loss": 0.5632,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998923821401394e-05,
+      "loss": 0.6607,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998918617130404e-05,
+      "loss": 0.6099,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989134003129405e-05,
+      "loss": 0.6195,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989081709490686e-05,
+      "loss": 0.6638,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989029290388543e-05,
+      "loss": 0.6374,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988976745823636e-05,
+      "loss": 0.6543,
+      "step": 1286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988924075796618e-05,
+      "loss": 0.6474,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988871280308154e-05,
+      "loss": 0.6,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998881835935891e-05,
+      "loss": 0.5902,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998876531294954e-05,
+      "loss": 0.5914,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988712141080727e-05,
+      "loss": 0.599,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988658843753126e-05,
+      "loss": 0.6626,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998860542096741e-05,
+      "loss": 0.5583,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988551872724252e-05,
+      "loss": 0.5641,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988498199024324e-05,
+      "loss": 0.6219,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99884443998683e-05,
+      "loss": 0.5395,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998839047525685e-05,
+      "loss": 0.6056,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988336425190663e-05,
+      "loss": 0.6365,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998828224967041e-05,
+      "loss": 0.6065,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988227948696766e-05,
+      "loss": 0.6321,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988173522270424e-05,
+      "loss": 0.6338,
+      "step": 1301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988118970392063e-05,
+      "loss": 0.5899,
+      "step": 1302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998806429306237e-05,
+      "loss": 0.5624,
+      "step": 1303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988009490282027e-05,
+      "loss": 0.5936,
+      "step": 1304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987954562051724e-05,
+      "loss": 0.6567,
+      "step": 1305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987899508372156e-05,
+      "loss": 0.6041,
+      "step": 1306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998784432924401e-05,
+      "loss": 0.6018,
+      "step": 1307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998778902466798e-05,
+      "loss": 0.6369,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998773359464476e-05,
+      "loss": 0.684,
+      "step": 1309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998767803917505e-05,
+      "loss": 0.6405,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987622358259536e-05,
+      "loss": 0.5975,
+      "step": 1311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987566551898935e-05,
+      "loss": 0.5744,
+      "step": 1312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987510620093934e-05,
+      "loss": 0.5612,
+      "step": 1313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998745456284524e-05,
+      "loss": 0.5933,
+      "step": 1314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987398380153556e-05,
+      "loss": 0.6444,
+      "step": 1315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987342072019592e-05,
+      "loss": 0.6437,
+      "step": 1316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987285638444053e-05,
+      "loss": 0.6763,
+      "step": 1317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987229079427645e-05,
+      "loss": 0.5699,
+      "step": 1318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998717239497108e-05,
+      "loss": 0.7229,
+      "step": 1319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998711558507507e-05,
+      "loss": 0.6668,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998705864974033e-05,
+      "loss": 0.6125,
+      "step": 1321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987001588967575e-05,
+      "loss": 0.5829,
+      "step": 1322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998694440275752e-05,
+      "loss": 0.622,
+      "step": 1323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986887091110887e-05,
+      "loss": 0.6523,
+      "step": 1324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998682965402839e-05,
+      "loss": 0.6285,
+      "step": 1325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986772091510753e-05,
+      "loss": 0.5497,
+      "step": 1326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986714403558695e-05,
+      "loss": 0.5855,
+      "step": 1327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998665659017295e-05,
+      "loss": 0.6265,
+      "step": 1328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998659865135424e-05,
+      "loss": 0.6611,
+      "step": 1329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986540587103288e-05,
+      "loss": 0.5867,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998648239742083e-05,
+      "loss": 0.6433,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998642408230759e-05,
+      "loss": 0.6111,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986365641764307e-05,
+      "loss": 0.6034,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986307075791713e-05,
+      "loss": 0.6804,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998624838439054e-05,
+      "loss": 0.6415,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986189567561533e-05,
+      "loss": 0.6474,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998613062530542e-05,
+      "loss": 0.6453,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986071557622947e-05,
+      "loss": 0.6116,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998601236451486e-05,
+      "loss": 0.6528,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985953045981895e-05,
+      "loss": 0.6341,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985893602024803e-05,
+      "loss": 0.6316,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985834032644325e-05,
+      "loss": 0.6222,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998577433784122e-05,
+      "loss": 0.6649,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985714517616222e-05,
+      "loss": 0.6438,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985654571970092e-05,
+      "loss": 0.6376,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985594500903583e-05,
+      "loss": 0.627,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985534304417448e-05,
+      "loss": 0.6554,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998547398251244e-05,
+      "loss": 0.5859,
+      "step": 1348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985413535189323e-05,
+      "loss": 0.5852,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985352962448852e-05,
+      "loss": 0.6432,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985292264291787e-05,
+      "loss": 0.6431,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985231440718894e-05,
+      "loss": 0.5762,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998517049173093e-05,
+      "loss": 0.6025,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998510941732867e-05,
+      "loss": 0.6486,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985048217512878e-05,
+      "loss": 0.6566,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984986892284318e-05,
+      "loss": 0.7019,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984925441643762e-05,
+      "loss": 0.6627,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984863865591986e-05,
+      "loss": 0.6572,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984802164129758e-05,
+      "loss": 0.6732,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998474033725786e-05,
+      "loss": 0.7076,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984678384977062e-05,
+      "loss": 0.5592,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984616307288143e-05,
+      "loss": 0.6523,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984554104191886e-05,
+      "loss": 0.6524,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984491775689067e-05,
+      "loss": 0.611,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984429321780475e-05,
+      "loss": 0.6127,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998436674246689e-05,
+      "loss": 0.5761,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99843040377491e-05,
+      "loss": 0.5762,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984241207627894e-05,
+      "loss": 0.6206,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998417825210406e-05,
+      "loss": 0.583,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984115171178384e-05,
+      "loss": 0.5605,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984051964851662e-05,
+      "loss": 0.6394,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998398863312469e-05,
+      "loss": 0.5599,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983925175998263e-05,
+      "loss": 0.6249,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983861593473174e-05,
+      "loss": 0.6261,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983797885550227e-05,
+      "loss": 0.6432,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998373405223022e-05,
+      "loss": 0.6724,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998367009351395e-05,
+      "loss": 0.5961,
+      "step": 1377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983606009402224e-05,
+      "loss": 0.6367,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998354179989585e-05,
+      "loss": 0.6625,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983477464995636e-05,
+      "loss": 0.5748,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983413004702383e-05,
+      "loss": 0.5098,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99833484190169e-05,
+      "loss": 0.5999,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983283707940006e-05,
+      "loss": 0.5718,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983218871472505e-05,
+      "loss": 0.5668,
+      "step": 1384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983153909615218e-05,
+      "loss": 0.6573,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998308882236896e-05,
+      "loss": 0.5833,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983023609734547e-05,
+      "loss": 0.6139,
+      "step": 1387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982958271712798e-05,
+      "loss": 0.573,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982892808304534e-05,
+      "loss": 0.5602,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982827219510575e-05,
+      "loss": 0.6354,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998276150533175e-05,
+      "loss": 0.682,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998269566576888e-05,
+      "loss": 0.6076,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982629700822792e-05,
+      "loss": 0.6142,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982563610494317e-05,
+      "loss": 0.4941,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982497394784285e-05,
+      "loss": 0.6501,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998243105369353e-05,
+      "loss": 0.5999,
+      "step": 1396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982364587222875e-05,
+      "loss": 0.6301,
+      "step": 1397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982297995373165e-05,
+      "loss": 0.6587,
+      "step": 1398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982231278145235e-05,
+      "loss": 0.5706,
+      "step": 1399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982164435539918e-05,
+      "loss": 0.6089,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982097467558058e-05,
+      "loss": 0.5982,
+      "step": 1401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982030374200492e-05,
+      "loss": 0.6753,
+      "step": 1402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998196315546807e-05,
+      "loss": 0.6007,
+      "step": 1403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981895811361627e-05,
+      "loss": 0.619,
+      "step": 1404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981828341882015e-05,
+      "loss": 0.5087,
+      "step": 1405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981760747030082e-05,
+      "loss": 0.6145,
+      "step": 1406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981693026806678e-05,
+      "loss": 0.6285,
+      "step": 1407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981625181212643e-05,
+      "loss": 0.6244,
+      "step": 1408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981557210248842e-05,
+      "loss": 0.6343,
+      "step": 1409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998148911391612e-05,
+      "loss": 0.5987,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998142089221534e-05,
+      "loss": 0.6072,
+      "step": 1411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998135254514735e-05,
+      "loss": 0.5842,
+      "step": 1412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981284072713016e-05,
+      "loss": 0.7175,
+      "step": 1413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981215474913195e-05,
+      "loss": 0.6321,
+      "step": 1414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981146751748744e-05,
+      "loss": 0.6302,
+      "step": 1415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981077903220537e-05,
+      "loss": 0.6543,
+      "step": 1416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981008929329428e-05,
+      "loss": 0.6037,
+      "step": 1417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998093983007629e-05,
+      "loss": 0.5911,
+      "step": 1418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980870605461988e-05,
+      "loss": 0.5979,
+      "step": 1419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980801255487393e-05,
+      "loss": 0.6138,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980731780153375e-05,
+      "loss": 0.5976,
+      "step": 1421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998066217946081e-05,
+      "loss": 0.5852,
+      "step": 1422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980592453410565e-05,
+      "loss": 0.6724,
+      "step": 1423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998052260200352e-05,
+      "loss": 0.6583,
+      "step": 1424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980452625240553e-05,
+      "loss": 0.6505,
+      "step": 1425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998038252312254e-05,
+      "loss": 0.6446,
+      "step": 1426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980312295650367e-05,
+      "loss": 0.5006,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998024194282491e-05,
+      "loss": 0.5578,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980171464647058e-05,
+      "loss": 0.5792,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998010086111769e-05,
+      "loss": 0.6765,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99800301322377e-05,
+      "loss": 0.6363,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979959278007966e-05,
+      "loss": 0.6409,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979888298429394e-05,
+      "loss": 0.6057,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997981719350286e-05,
+      "loss": 0.6631,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979745963229263e-05,
+      "loss": 0.6493,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99796746076095e-05,
+      "loss": 0.6064,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979603126644466e-05,
+      "loss": 0.6068,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979531520335052e-05,
+      "loss": 0.5986,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997945978868217e-05,
+      "loss": 0.5951,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997938793168671e-05,
+      "loss": 0.6493,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997931594934958e-05,
+      "loss": 0.5897,
+      "step": 1441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979243841671684e-05,
+      "loss": 0.6492,
+      "step": 1442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979171608653926e-05,
+      "loss": 0.636,
+      "step": 1443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997909925029721e-05,
+      "loss": 0.6443,
+      "step": 1444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979026766602453e-05,
+      "loss": 0.5733,
+      "step": 1445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978954157570558e-05,
+      "loss": 0.6375,
+      "step": 1446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978881423202442e-05,
+      "loss": 0.6399,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978808563499013e-05,
+      "loss": 0.6212,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978735578461193e-05,
+      "loss": 0.6677,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978662468089897e-05,
+      "loss": 0.6136,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.6437,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978515871350538e-05,
+      "loss": 0.5964,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978442384984325e-05,
+      "loss": 0.6402,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997836877328831e-05,
+      "loss": 0.5985,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997829503626343e-05,
+      "loss": 0.6548,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978221173910605e-05,
+      "loss": 0.6288,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997814718623076e-05,
+      "loss": 0.6079,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997807307322483e-05,
+      "loss": 0.6039,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977998834893744e-05,
+      "loss": 0.641,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997792447123843e-05,
+      "loss": 0.5831,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997784998225983e-05,
+      "loss": 0.6813,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977775367958873e-05,
+      "loss": 0.6974,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99777006283365e-05,
+      "loss": 0.638,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977625763393646e-05,
+      "loss": 0.5756,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977550773131257e-05,
+      "loss": 0.6789,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997747565755027e-05,
+      "loss": 0.6397,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997740041665163e-05,
+      "loss": 0.7193,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997732505043628e-05,
+      "loss": 0.6842,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977249558905173e-05,
+      "loss": 0.622,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977173942059248e-05,
+      "loss": 0.631,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977098199899464e-05,
+      "loss": 0.5957,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977022332426763e-05,
+      "loss": 0.5983,
+      "step": 1472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976946339642107e-05,
+      "loss": 0.5641,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976870221546444e-05,
+      "loss": 0.6524,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997679397814073e-05,
+      "loss": 0.6337,
+      "step": 1475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997671760942593e-05,
+      "loss": 0.6073,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976641115402994e-05,
+      "loss": 0.6379,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976564496072888e-05,
+      "loss": 0.6407,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976487751436574e-05,
+      "loss": 0.5699,
+      "step": 1479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976410881495013e-05,
+      "loss": 0.7219,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976333886249178e-05,
+      "loss": 0.5791,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976256765700027e-05,
+      "loss": 0.6289,
+      "step": 1482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976179519848527e-05,
+      "loss": 0.5628,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997610214869566e-05,
+      "loss": 0.5832,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976024652242392e-05,
+      "loss": 0.6051,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975947030489694e-05,
+      "loss": 0.6446,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997586928343854e-05,
+      "loss": 0.5786,
+      "step": 1487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975791411089912e-05,
+      "loss": 0.662,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975713413444782e-05,
+      "loss": 0.5673,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975635290504137e-05,
+      "loss": 0.6185,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975557042268955e-05,
+      "loss": 0.6151,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975478668740215e-05,
+      "loss": 0.6105,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99754001699189e-05,
+      "loss": 0.6513,
+      "step": 1493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975321545806004e-05,
+      "loss": 0.565,
+      "step": 1494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975242796402515e-05,
+      "loss": 0.5624,
+      "step": 1495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975163921709412e-05,
+      "loss": 0.6265,
+      "step": 1496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975084921727698e-05,
+      "loss": 0.6011,
+      "step": 1497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997500579645835e-05,
+      "loss": 0.5878,
+      "step": 1498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974926545902376e-05,
+      "loss": 0.6128,
+      "step": 1499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974847170060766e-05,
+      "loss": 0.5929,
+      "step": 1500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974767668934513e-05,
+      "loss": 0.6554,
+      "step": 1501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974688042524623e-05,
+      "loss": 0.5855,
+      "step": 1502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997460829083209e-05,
+      "loss": 0.5375,
+      "step": 1503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997452841385792e-05,
+      "loss": 0.6294,
+      "step": 1504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974448411603116e-05,
+      "loss": 0.6201,
+      "step": 1505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997436828406868e-05,
+      "loss": 0.7028,
+      "step": 1506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997428803125562e-05,
+      "loss": 0.5956,
+      "step": 1507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974207653164943e-05,
+      "loss": 0.6245,
+      "step": 1508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974127149797658e-05,
+      "loss": 0.6156,
+      "step": 1509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974046521154776e-05,
+      "loss": 0.6026,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973965767237317e-05,
+      "loss": 0.6275,
+      "step": 1511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973884888046284e-05,
+      "loss": 0.6061,
+      "step": 1512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973803883582703e-05,
+      "loss": 0.5374,
+      "step": 1513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997372275384758e-05,
+      "loss": 0.6271,
+      "step": 1514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973641498841948e-05,
+      "loss": 0.6137,
+      "step": 1515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973560118566813e-05,
+      "loss": 0.6197,
+      "step": 1516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997347861302321e-05,
+      "loss": 0.6132,
+      "step": 1517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973396982212155e-05,
+      "loss": 0.5906,
+      "step": 1518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973315226134678e-05,
+      "loss": 0.6321,
+      "step": 1519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99732333447918e-05,
+      "loss": 0.6502,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997315133818455e-05,
+      "loss": 0.663,
+      "step": 1521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973069206313965e-05,
+      "loss": 0.6335,
+      "step": 1522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972986949181074e-05,
+      "loss": 0.5516,
+      "step": 1523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972904566786903e-05,
+      "loss": 0.6039,
+      "step": 1524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972822059132495e-05,
+      "loss": 0.6727,
+      "step": 1525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972739426218884e-05,
+      "loss": 0.5935,
+      "step": 1526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972656668047105e-05,
+      "loss": 0.6215,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99725737846182e-05,
+      "loss": 0.6056,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972490775933213e-05,
+      "loss": 0.6305,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997240764199318e-05,
+      "loss": 0.6361,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997232438279915e-05,
+      "loss": 0.6171,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997224099835217e-05,
+      "loss": 0.6393,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972157488653277e-05,
+      "loss": 0.5483,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972073853703533e-05,
+      "loss": 0.6253,
+      "step": 1534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997199009350398e-05,
+      "loss": 0.6358,
+      "step": 1535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971906208055677e-05,
+      "loss": 0.6066,
+      "step": 1536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997182219735967e-05,
+      "loss": 0.5859,
+      "step": 1537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997173806141702e-05,
+      "loss": 0.5504,
+      "step": 1538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971653800228782e-05,
+      "loss": 0.5741,
+      "step": 1539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997156941379601e-05,
+      "loss": 0.5773,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997148490211977e-05,
+      "loss": 0.5887,
+      "step": 1541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997140026520112e-05,
+      "loss": 0.5777,
+      "step": 1542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971315503041125e-05,
+      "loss": 0.7153,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971230615640847e-05,
+      "loss": 0.5843,
+      "step": 1544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971145603001358e-05,
+      "loss": 0.5485,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971060465123718e-05,
+      "loss": 0.5958,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970975202009003e-05,
+      "loss": 0.653,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997088981365828e-05,
+      "loss": 0.612,
+      "step": 1548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970804300072622e-05,
+      "loss": 0.6401,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970718661253107e-05,
+      "loss": 0.6337,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970632897200804e-05,
+      "loss": 0.5951,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970547007916793e-05,
+      "loss": 0.5861,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970460993402153e-05,
+      "loss": 0.6321,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970374853657965e-05,
+      "loss": 0.5726,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970288588685308e-05,
+      "loss": 0.625,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970202198485274e-05,
+      "loss": 0.6258,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970115683058937e-05,
+      "loss": 0.6057,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970029042407392e-05,
+      "loss": 0.6482,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969942276531722e-05,
+      "loss": 0.624,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969855385433015e-05,
+      "loss": 0.5408,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969768369112373e-05,
+      "loss": 0.6117,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996968122757088e-05,
+      "loss": 0.5927,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996959396080963e-05,
+      "loss": 0.632,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996950656882972e-05,
+      "loss": 0.6223,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969419051632256e-05,
+      "loss": 0.5748,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969331409218324e-05,
+      "loss": 0.591,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969243641589032e-05,
+      "loss": 0.6097,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969155748745482e-05,
+      "loss": 0.6563,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996906773068878e-05,
+      "loss": 0.5736,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968979587420026e-05,
+      "loss": 0.6524,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968891318940332e-05,
+      "loss": 0.6412,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968802925250802e-05,
+      "loss": 0.5951,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968714406352553e-05,
+      "loss": 0.5705,
+      "step": 1573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968625762246693e-05,
+      "loss": 0.5586,
+      "step": 1574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996853699293433e-05,
+      "loss": 0.6372,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968448098416587e-05,
+      "loss": 0.5859,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996835907869458e-05,
+      "loss": 0.6454,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996826993376942e-05,
+      "loss": 0.6629,
+      "step": 1578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968180663642235e-05,
+      "loss": 0.6513,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968091268314137e-05,
+      "loss": 0.6114,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996800174778626e-05,
+      "loss": 0.6221,
+      "step": 1581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967912102059723e-05,
+      "loss": 0.6218,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967822331135648e-05,
+      "loss": 0.6223,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996773243501517e-05,
+      "loss": 0.6223,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967642413699413e-05,
+      "loss": 0.585,
+      "step": 1585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967552267189506e-05,
+      "loss": 0.6115,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996746199548659e-05,
+      "loss": 0.6333,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996737159859179e-05,
+      "loss": 0.6255,
+      "step": 1588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996728107650624e-05,
+      "loss": 0.6573,
+      "step": 1589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996719042923109e-05,
+      "loss": 0.6189,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967099656767465e-05,
+      "loss": 0.6403,
+      "step": 1591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996700875911651e-05,
+      "loss": 0.6259,
+      "step": 1592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966917736279368e-05,
+      "loss": 0.5865,
+      "step": 1593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996682658825718e-05,
+      "loss": 0.7139,
+      "step": 1594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966735315051095e-05,
+      "loss": 0.6229,
+      "step": 1595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996664391666225e-05,
+      "loss": 0.5802,
+      "step": 1596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966552393091804e-05,
+      "loss": 0.6383,
+      "step": 1597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966460744340902e-05,
+      "loss": 0.6573,
+      "step": 1598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996636897041069e-05,
+      "loss": 0.5006,
+      "step": 1599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966277071302335e-05,
+      "loss": 0.6318,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966185047016974e-05,
+      "loss": 0.6654,
+      "step": 1601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966092897555772e-05,
+      "loss": 0.5861,
+      "step": 1602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966000622919886e-05,
+      "loss": 0.6372,
+      "step": 1603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965908223110474e-05,
+      "loss": 0.6361,
+      "step": 1604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965815698128698e-05,
+      "loss": 0.633,
+      "step": 1605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965723047975713e-05,
+      "loss": 0.6218,
+      "step": 1606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965630272652695e-05,
+      "loss": 0.6305,
+      "step": 1607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99655373721608e-05,
+      "loss": 0.6248,
+      "step": 1608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965444346501196e-05,
+      "loss": 0.7028,
+      "step": 1609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965351195675056e-05,
+      "loss": 0.5889,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996525791968354e-05,
+      "loss": 0.6179,
+      "step": 1611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965164518527835e-05,
+      "loss": 0.6607,
+      "step": 1612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965070992209103e-05,
+      "loss": 0.6083,
+      "step": 1613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964977340728524e-05,
+      "loss": 0.6533,
+      "step": 1614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964883564087266e-05,
+      "loss": 0.6277,
+      "step": 1615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964789662286512e-05,
+      "loss": 0.6094,
+      "step": 1616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964695635327445e-05,
+      "loss": 0.6937,
+      "step": 1617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996460148321124e-05,
+      "loss": 0.5832,
+      "step": 1618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964507205939084e-05,
+      "loss": 0.5943,
+      "step": 1619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996441280351216e-05,
+      "loss": 0.5739,
+      "step": 1620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964318275931652e-05,
+      "loss": 0.5967,
+      "step": 1621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964223623198752e-05,
+      "loss": 0.6226,
+      "step": 1622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996412884531464e-05,
+      "loss": 0.5994,
+      "step": 1623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964033942280516e-05,
+      "loss": 0.5792,
+      "step": 1624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996393891409756e-05,
+      "loss": 0.6926,
+      "step": 1625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963843760766983e-05,
+      "loss": 0.6702,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963748482289963e-05,
+      "loss": 0.6421,
+      "step": 1627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996365307866771e-05,
+      "loss": 0.6042,
+      "step": 1628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996355754990141e-05,
+      "loss": 0.5986,
+      "step": 1629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963461895992274e-05,
+      "loss": 0.5982,
+      "step": 1630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963366116941497e-05,
+      "loss": 0.5955,
+      "step": 1631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963270212750286e-05,
+      "loss": 0.5943,
+      "step": 1632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996317418341984e-05,
+      "loss": 0.5917,
+      "step": 1633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963078028951368e-05,
+      "loss": 0.5531,
+      "step": 1634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996298174934608e-05,
+      "loss": 0.6294,
+      "step": 1635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996288534460518e-05,
+      "loss": 0.6665,
+      "step": 1636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962788814729884e-05,
+      "loss": 0.572,
+      "step": 1637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962692159721407e-05,
+      "loss": 0.5805,
+      "step": 1638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962595379580953e-05,
+      "loss": 0.6351,
+      "step": 1639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996249847430974e-05,
+      "loss": 0.5504,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962401443908995e-05,
+      "loss": 0.6516,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962304288379927e-05,
+      "loss": 0.5749,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996220700772376e-05,
+      "loss": 0.6069,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962109601941715e-05,
+      "loss": 0.6362,
+      "step": 1644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962012071035015e-05,
+      "loss": 0.5953,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961914415004885e-05,
+      "loss": 0.6293,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961816633852552e-05,
+      "loss": 0.6335,
+      "step": 1647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961718727579247e-05,
+      "loss": 0.6395,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961620696186195e-05,
+      "loss": 0.5741,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961522539674627e-05,
+      "loss": 0.6849,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996142425804578e-05,
+      "loss": 0.557,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996132585130089e-05,
+      "loss": 0.7227,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996122731944118e-05,
+      "loss": 0.6211,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961128662467905e-05,
+      "loss": 0.6743,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996102988038229e-05,
+      "loss": 0.5592,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996093097318559e-05,
+      "loss": 0.6228,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960831940879033e-05,
+      "loss": 0.5934,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996073278346387e-05,
+      "loss": 0.6306,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960633500941347e-05,
+      "loss": 0.5919,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996053409331271e-05,
+      "loss": 0.6548,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960434560579204e-05,
+      "loss": 0.6364,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960334902742083e-05,
+      "loss": 0.677,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960235119802597e-05,
+      "loss": 0.5995,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960135211762002e-05,
+      "loss": 0.5738,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996003517862155e-05,
+      "loss": 0.5741,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99599350203825e-05,
+      "loss": 0.6461,
+      "step": 1666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959834737046107e-05,
+      "loss": 0.5809,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959734328613633e-05,
+      "loss": 0.6517,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959633795086332e-05,
+      "loss": 0.6129,
+      "step": 1669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995953313646548e-05,
+      "loss": 0.5856,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995943235275233e-05,
+      "loss": 0.5446,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959331443948154e-05,
+      "loss": 0.6271,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995923041005422e-05,
+      "loss": 0.5503,
+      "step": 1673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959129251071786e-05,
+      "loss": 0.6703,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959027967002137e-05,
+      "loss": 0.6578,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995892655784654e-05,
+      "loss": 0.5928,
+      "step": 1676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995882502360626e-05,
+      "loss": 0.5954,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958723364282588e-05,
+      "loss": 0.5947,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995862157987679e-05,
+      "loss": 0.569,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995851967039015e-05,
+      "loss": 0.624,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995841763582394e-05,
+      "loss": 0.6032,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995831547617945e-05,
+      "loss": 0.6746,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958213191457958e-05,
+      "loss": 0.6676,
+      "step": 1683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958110781660752e-05,
+      "loss": 0.5616,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958008246789117e-05,
+      "loss": 0.619,
+      "step": 1685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995790558684434e-05,
+      "loss": 0.6814,
+      "step": 1686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995780280182771e-05,
+      "loss": 0.5951,
+      "step": 1687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995769989174052e-05,
+      "loss": 0.6282,
+      "step": 1688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957596856584063e-05,
+      "loss": 0.6648,
+      "step": 1689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995749369635963e-05,
+      "loss": 0.5845,
+      "step": 1690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995739041106852e-05,
+      "loss": 0.6742,
+      "step": 1691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957287000712027e-05,
+      "loss": 0.6132,
+      "step": 1692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957183465291454e-05,
+      "loss": 0.6,
+      "step": 1693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99570798048081e-05,
+      "loss": 0.6471,
+      "step": 1694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995697601926326e-05,
+      "loss": 0.5571,
+      "step": 1695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956872108658246e-05,
+      "loss": 0.64,
+      "step": 1696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956768072994363e-05,
+      "loss": 0.5867,
+      "step": 1697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956663912272916e-05,
+      "loss": 0.5616,
+      "step": 1698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995655962649521e-05,
+      "loss": 0.7113,
+      "step": 1699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995645521566256e-05,
+      "loss": 0.6079,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956350679776274e-05,
+      "loss": 0.6191,
+      "step": 1701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956246018837663e-05,
+      "loss": 0.6459,
+      "step": 1702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995614123284805e-05,
+      "loss": 0.6461,
+      "step": 1703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995603632180874e-05,
+      "loss": 0.6058,
+      "step": 1704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995593128572106e-05,
+      "loss": 0.613,
+      "step": 1705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955826124586323e-05,
+      "loss": 0.6297,
+      "step": 1706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955720838405853e-05,
+      "loss": 0.6217,
+      "step": 1707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995561542718097e-05,
+      "loss": 0.6748,
+      "step": 1708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955509890913e-05,
+      "loss": 0.6382,
+      "step": 1709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995540422960327e-05,
+      "loss": 0.6597,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955298443253107e-05,
+      "loss": 0.6008,
+      "step": 1711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955192531863836e-05,
+      "loss": 0.5921,
+      "step": 1712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955086495436787e-05,
+      "loss": 0.5312,
+      "step": 1713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954980333973297e-05,
+      "loss": 0.6075,
+      "step": 1714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954874047474696e-05,
+      "loss": 0.4977,
+      "step": 1715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954767635942322e-05,
+      "loss": 0.6094,
+      "step": 1716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99546610993775e-05,
+      "loss": 0.6413,
+      "step": 1717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954554437781587e-05,
+      "loss": 0.5973,
+      "step": 1718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995444765115591e-05,
+      "loss": 0.5802,
+      "step": 1719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995434073950181e-05,
+      "loss": 0.568,
+      "step": 1720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954233702820634e-05,
+      "loss": 0.5926,
+      "step": 1721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954126541113727e-05,
+      "loss": 0.6426,
+      "step": 1722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995401925438243e-05,
+      "loss": 0.6306,
+      "step": 1723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953911842628097e-05,
+      "loss": 0.6027,
+      "step": 1724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995380430585207e-05,
+      "loss": 0.5991,
+      "step": 1725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995369664405571e-05,
+      "loss": 0.611,
+      "step": 1726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953588857240357e-05,
+      "loss": 0.6723,
+      "step": 1727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995348094540737e-05,
+      "loss": 0.6338,
+      "step": 1728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953372908558108e-05,
+      "loss": 0.6042,
+      "step": 1729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953264746693925e-05,
+      "loss": 0.6438,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995315645981618e-05,
+      "loss": 0.6245,
+      "step": 1731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995304804792623e-05,
+      "loss": 0.5464,
+      "step": 1732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952939511025436e-05,
+      "loss": 0.6256,
+      "step": 1733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952830849115166e-05,
+      "loss": 0.6492,
+      "step": 1734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995272206219679e-05,
+      "loss": 0.6012,
+      "step": 1735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995261315027166e-05,
+      "loss": 0.5666,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952504113341153e-05,
+      "loss": 0.6151,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995239495140664e-05,
+      "loss": 0.6754,
+      "step": 1738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952285664469488e-05,
+      "loss": 0.6128,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995217625253107e-05,
+      "loss": 0.6585,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995206671559276e-05,
+      "loss": 0.6886,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951957053655936e-05,
+      "loss": 0.5701,
+      "step": 1742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.5978,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951737354792247e-05,
+      "loss": 0.6195,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995162731786815e-05,
+      "loss": 0.645,
+      "step": 1745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995151715595105e-05,
+      "loss": 0.6131,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951406869042343e-05,
+      "loss": 0.628,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951296457143403e-05,
+      "loss": 0.6125,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951185920255627e-05,
+      "loss": 0.5573,
+      "step": 1749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951075258380395e-05,
+      "loss": 0.6232,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950964471519102e-05,
+      "loss": 0.6859,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950853559673138e-05,
+      "loss": 0.5795,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950742522843894e-05,
+      "loss": 0.639,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950631361032767e-05,
+      "loss": 0.6691,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950520074241157e-05,
+      "loss": 0.6226,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995040866247045e-05,
+      "loss": 0.6303,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995029712572206e-05,
+      "loss": 0.5782,
+      "step": 1757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995018546399738e-05,
+      "loss": 0.62,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995007367729781e-05,
+      "loss": 0.5651,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994996176562476e-05,
+      "loss": 0.6745,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994984972897963e-05,
+      "loss": 0.5149,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949737567363835e-05,
+      "loss": 0.6133,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994962528077878e-05,
+      "loss": 0.5408,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949512869225867e-05,
+      "loss": 0.6094,
+      "step": 1764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949400332706518e-05,
+      "loss": 0.618,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949287671222145e-05,
+      "loss": 0.6199,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994917488477416e-05,
+      "loss": 0.6195,
+      "step": 1767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994906197336398e-05,
+      "loss": 0.5921,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948948936993026e-05,
+      "loss": 0.6641,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948835775662717e-05,
+      "loss": 0.6055,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948722489374472e-05,
+      "loss": 0.5165,
+      "step": 1771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948609078129718e-05,
+      "loss": 0.5439,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948495541929874e-05,
+      "loss": 0.5882,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994838188077637e-05,
+      "loss": 0.6314,
+      "step": 1774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948268094670632e-05,
+      "loss": 0.6036,
+      "step": 1775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948154183614094e-05,
+      "loss": 0.68,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948040147608175e-05,
+      "loss": 0.6293,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994792598665432e-05,
+      "loss": 0.6333,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947811700753955e-05,
+      "loss": 0.5868,
+      "step": 1779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947697289908516e-05,
+      "loss": 0.5685,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947582754119443e-05,
+      "loss": 0.6746,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947468093388175e-05,
+      "loss": 0.6637,
+      "step": 1782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994735330771615e-05,
+      "loss": 0.6243,
+      "step": 1783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994723839710481e-05,
+      "loss": 0.6431,
+      "step": 1784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947123361555597e-05,
+      "loss": 0.5477,
+      "step": 1785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994700820106996e-05,
+      "loss": 0.6526,
+      "step": 1786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946892915649343e-05,
+      "loss": 0.6197,
+      "step": 1787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946777505295195e-05,
+      "loss": 0.565,
+      "step": 1788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994666197000896e-05,
+      "loss": 0.6048,
+      "step": 1789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946546309792097e-05,
+      "loss": 0.605,
+      "step": 1790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946430524646056e-05,
+      "loss": 0.6307,
+      "step": 1791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994631461457229e-05,
+      "loss": 0.5547,
+      "step": 1792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946198579572255e-05,
+      "loss": 0.6343,
+      "step": 1793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994608241964741e-05,
+      "loss": 0.6562,
+      "step": 1794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945966134799207e-05,
+      "loss": 0.6337,
+      "step": 1795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994584972502912e-05,
+      "loss": 0.6452,
+      "step": 1796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945733190338598e-05,
+      "loss": 0.5921,
+      "step": 1797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994561653072911e-05,
+      "loss": 0.6008,
+      "step": 1798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945499746202123e-05,
+      "loss": 0.544,
+      "step": 1799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945382836759103e-05,
+      "loss": 0.6639,
+      "step": 1800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994526580240152e-05,
+      "loss": 0.6876,
+      "step": 1801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945148643130837e-05,
+      "loss": 0.5933,
+      "step": 1802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945031358948532e-05,
+      "loss": 0.556,
+      "step": 1803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944913949856074e-05,
+      "loss": 0.5872,
+      "step": 1804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944796415854944e-05,
+      "loss": 0.6159,
+      "step": 1805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994467875694661e-05,
+      "loss": 0.6208,
+      "step": 1806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944560973132557e-05,
+      "loss": 0.5168,
+      "step": 1807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994444306441426e-05,
+      "loss": 0.6527,
+      "step": 1808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944325030793198e-05,
+      "loss": 0.6207,
+      "step": 1809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994420687227086e-05,
+      "loss": 0.68,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944088588848728e-05,
+      "loss": 0.7005,
+      "step": 1811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943970180528283e-05,
+      "loss": 0.6429,
+      "step": 1812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943851647311018e-05,
+      "loss": 0.6599,
+      "step": 1813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994373298919842e-05,
+      "loss": 0.6755,
+      "step": 1814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994361420619198e-05,
+      "loss": 0.5582,
+      "step": 1815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943495298293182e-05,
+      "loss": 0.6008,
+      "step": 1816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994337626550353e-05,
+      "loss": 0.633,
+      "step": 1817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943257107824518e-05,
+      "loss": 0.6677,
+      "step": 1818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943137825257634e-05,
+      "loss": 0.6939,
+      "step": 1819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943018417804386e-05,
+      "loss": 0.5587,
+      "step": 1820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942898885466267e-05,
+      "loss": 0.6646,
+      "step": 1821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942779228244785e-05,
+      "loss": 0.6011,
+      "step": 1822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942659446141438e-05,
+      "loss": 0.6273,
+      "step": 1823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942539539157732e-05,
+      "loss": 0.5264,
+      "step": 1824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942419507295174e-05,
+      "loss": 0.5712,
+      "step": 1825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942299350555262e-05,
+      "loss": 0.6248,
+      "step": 1826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994217906893952e-05,
+      "loss": 0.6286,
+      "step": 1827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942058662449454e-05,
+      "loss": 0.605,
+      "step": 1828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994193813108657e-05,
+      "loss": 0.5819,
+      "step": 1829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941817474852385e-05,
+      "loss": 0.5599,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994169669374842e-05,
+      "loss": 0.538,
+      "step": 1831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941575787776186e-05,
+      "loss": 0.5615,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941454756937203e-05,
+      "loss": 0.5489,
+      "step": 1833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994133360123299e-05,
+      "loss": 0.5923,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941212320665072e-05,
+      "loss": 0.7131,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941090915234973e-05,
+      "loss": 0.6504,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940969384944212e-05,
+      "loss": 0.6705,
+      "step": 1837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994084772979432e-05,
+      "loss": 0.6353,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994072594978682e-05,
+      "loss": 0.5975,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940604044923254e-05,
+      "loss": 0.66,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940482015205137e-05,
+      "loss": 0.597,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994035986063401e-05,
+      "loss": 0.618,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940237581211407e-05,
+      "loss": 0.6069,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940115176938867e-05,
+      "loss": 0.6145,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939992647817917e-05,
+      "loss": 0.5821,
+      "step": 1845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993986999385011e-05,
+      "loss": 0.5689,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939747215036973e-05,
+      "loss": 0.6342,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939624311380055e-05,
+      "loss": 0.5789,
+      "step": 1848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939501282880904e-05,
+      "loss": 0.6343,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939378129541056e-05,
+      "loss": 0.558,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939254851362063e-05,
+      "loss": 0.576,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993913144834547e-05,
+      "loss": 0.5403,
+      "step": 1852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939007920492834e-05,
+      "loss": 0.6473,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938884267805698e-05,
+      "loss": 0.6444,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993876049028562e-05,
+      "loss": 0.5368,
+      "step": 1855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993863658793416e-05,
+      "loss": 0.6413,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993851256075286e-05,
+      "loss": 0.6148,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938388408743287e-05,
+      "loss": 0.5867,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938264131906998e-05,
+      "loss": 0.5905,
+      "step": 1859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938139730245557e-05,
+      "loss": 0.5773,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938015203760523e-05,
+      "loss": 0.62,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937890552453462e-05,
+      "loss": 0.5951,
+      "step": 1862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993776577632594e-05,
+      "loss": 0.5943,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993764087537952e-05,
+      "loss": 0.5554,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993751584961578e-05,
+      "loss": 0.6003,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993739069903628e-05,
+      "loss": 0.5649,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937265423642597e-05,
+      "loss": 0.6203,
+      "step": 1867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937140023436306e-05,
+      "loss": 0.6194,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937014498418978e-05,
+      "loss": 0.6215,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936888848592192e-05,
+      "loss": 0.6085,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936763073957524e-05,
+      "loss": 0.6587,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936637174516552e-05,
+      "loss": 0.6238,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936511150270866e-05,
+      "loss": 0.6405,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993638500122204e-05,
+      "loss": 0.6604,
+      "step": 1874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936258727371667e-05,
+      "loss": 0.6616,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936132328721325e-05,
+      "loss": 0.5937,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936005805272604e-05,
+      "loss": 0.6088,
+      "step": 1877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935879157027094e-05,
+      "loss": 0.6975,
+      "step": 1878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935752383986386e-05,
+      "loss": 0.5482,
+      "step": 1879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993562548615207e-05,
+      "loss": 0.5561,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935498463525743e-05,
+      "loss": 0.6144,
+      "step": 1881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935371316108996e-05,
+      "loss": 0.5808,
+      "step": 1882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993524404390343e-05,
+      "loss": 0.6639,
+      "step": 1883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935116646910645e-05,
+      "loss": 0.6072,
+      "step": 1884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934989125132236e-05,
+      "loss": 0.6512,
+      "step": 1885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934861478569808e-05,
+      "loss": 0.5778,
+      "step": 1886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934733707224963e-05,
+      "loss": 0.5957,
+      "step": 1887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934605811099307e-05,
+      "loss": 0.5452,
+      "step": 1888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934477790194445e-05,
+      "loss": 0.6081,
+      "step": 1889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934349644511988e-05,
+      "loss": 0.5852,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934221374053538e-05,
+      "loss": 0.5981,
+      "step": 1891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934092978820714e-05,
+      "loss": 0.5989,
+      "step": 1892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933964458815126e-05,
+      "loss": 0.5541,
+      "step": 1893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933835814038386e-05,
+      "loss": 0.6415,
+      "step": 1894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933707044492118e-05,
+      "loss": 0.5296,
+      "step": 1895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933578150177927e-05,
+      "loss": 0.5965,
+      "step": 1896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993344913109744e-05,
+      "loss": 0.6356,
+      "step": 1897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933319987252275e-05,
+      "loss": 0.5805,
+      "step": 1898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933190718644054e-05,
+      "loss": 0.6033,
+      "step": 1899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933061325274405e-05,
+      "loss": 0.5991,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932931807144948e-05,
+      "loss": 0.5882,
+      "step": 1901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993280216425731e-05,
+      "loss": 0.6626,
+      "step": 1902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932672396613118e-05,
+      "loss": 0.6476,
+      "step": 1903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932542504214008e-05,
+      "loss": 0.594,
+      "step": 1904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932412487061606e-05,
+      "loss": 0.5537,
+      "step": 1905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932282345157547e-05,
+      "loss": 0.6332,
+      "step": 1906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932152078503465e-05,
+      "loss": 0.5992,
+      "step": 1907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932021687101e-05,
+      "loss": 0.6387,
+      "step": 1908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993189117095178e-05,
+      "loss": 0.5126,
+      "step": 1909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993176053005746e-05,
+      "loss": 0.6256,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931629764419666e-05,
+      "loss": 0.5333,
+      "step": 1911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931498874040046e-05,
+      "loss": 0.6141,
+      "step": 1912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993136785892024e-05,
+      "loss": 0.6205,
+      "step": 1913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931236719061903e-05,
+      "loss": 0.6287,
+      "step": 1914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931105454466678e-05,
+      "loss": 0.6154,
+      "step": 1915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993097406513621e-05,
+      "loss": 0.6694,
+      "step": 1916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930842551072148e-05,
+      "loss": 0.6367,
+      "step": 1917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993071091227615e-05,
+      "loss": 0.587,
+      "step": 1918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993057914874987e-05,
+      "loss": 0.6362,
+      "step": 1919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930447260494957e-05,
+      "loss": 0.6015,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993031524751307e-05,
+      "loss": 0.6444,
+      "step": 1921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930183109805866e-05,
+      "loss": 0.5939,
+      "step": 1922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993005084737501e-05,
+      "loss": 0.6092,
+      "step": 1923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929918460222155e-05,
+      "loss": 0.583,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992978594834897e-05,
+      "loss": 0.5842,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929653311757115e-05,
+      "loss": 0.5448,
+      "step": 1926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992952055044826e-05,
+      "loss": 0.6603,
+      "step": 1927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929387664424072e-05,
+      "loss": 0.6166,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929254653686215e-05,
+      "loss": 0.5606,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992912151823637e-05,
+      "loss": 0.681,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99289882580762e-05,
+      "loss": 0.6482,
+      "step": 1931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992885487320738e-05,
+      "loss": 0.59,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928721363631585e-05,
+      "loss": 0.6415,
+      "step": 1933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928587729350495e-05,
+      "loss": 0.5773,
+      "step": 1934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928453970365788e-05,
+      "loss": 0.6153,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928320086679146e-05,
+      "loss": 0.6,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928186078292243e-05,
+      "loss": 0.5506,
+      "step": 1937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992805194520677e-05,
+      "loss": 0.6501,
+      "step": 1938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927917687424404e-05,
+      "loss": 0.5689,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927783304946836e-05,
+      "loss": 0.6033,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992764879777576e-05,
+      "loss": 0.6339,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927514165912853e-05,
+      "loss": 0.6386,
+      "step": 1942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927379409359816e-05,
+      "loss": 0.6772,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927244528118336e-05,
+      "loss": 0.6279,
+      "step": 1944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992710952219011e-05,
+      "loss": 0.5653,
+      "step": 1945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992697439157683e-05,
+      "loss": 0.6583,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926839136280195e-05,
+      "loss": 0.5783,
+      "step": 1947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926703756301907e-05,
+      "loss": 0.623,
+      "step": 1948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926568251643665e-05,
+      "loss": 0.6397,
+      "step": 1949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992643262230717e-05,
+      "loss": 0.6286,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926296868294123e-05,
+      "loss": 0.5877,
+      "step": 1951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992616098960623e-05,
+      "loss": 0.6108,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926024986245202e-05,
+      "loss": 0.6632,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925888858212744e-05,
+      "loss": 0.6718,
+      "step": 1954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925752605510566e-05,
+      "loss": 0.6211,
+      "step": 1955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925616228140377e-05,
+      "loss": 0.5859,
+      "step": 1956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925479726103893e-05,
+      "loss": 0.5895,
+      "step": 1957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925343099402828e-05,
+      "loss": 0.5317,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99252063480389e-05,
+      "loss": 0.6531,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992506947201382e-05,
+      "loss": 0.5264,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924932471329316e-05,
+      "loss": 0.6377,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924795345987103e-05,
+      "loss": 0.5666,
+      "step": 1962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924658095988902e-05,
+      "loss": 0.5684,
+      "step": 1963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924520721336443e-05,
+      "loss": 0.5847,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924383222031447e-05,
+      "loss": 0.573,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992424559807564e-05,
+      "loss": 0.5907,
+      "step": 1966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924107849470756e-05,
+      "loss": 0.6724,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992396997621852e-05,
+      "loss": 0.611,
+      "step": 1968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992383197832066e-05,
+      "loss": 0.6567,
+      "step": 1969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992369385577892e-05,
+      "loss": 0.5866,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992355560859503e-05,
+      "loss": 0.6441,
+      "step": 1971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923417236770726e-05,
+      "loss": 0.504,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923278740307744e-05,
+      "loss": 0.6277,
+      "step": 1973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923140119207827e-05,
+      "loss": 0.5896,
+      "step": 1974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992300137347271e-05,
+      "loss": 0.5529,
+      "step": 1975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922862503104145e-05,
+      "loss": 0.6576,
+      "step": 1976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922723508103868e-05,
+      "loss": 0.6597,
+      "step": 1977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992258438847363e-05,
+      "loss": 0.6628,
+      "step": 1978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922445144215173e-05,
+      "loss": 0.5909,
+      "step": 1979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922305775330254e-05,
+      "loss": 0.6023,
+      "step": 1980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922166281820615e-05,
+      "loss": 0.5968,
+      "step": 1981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922026663688015e-05,
+      "loss": 0.6144,
+      "step": 1982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921886920934202e-05,
+      "loss": 0.667,
+      "step": 1983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921747053560934e-05,
+      "loss": 0.5753,
+      "step": 1984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921607061569966e-05,
+      "loss": 0.6054,
+      "step": 1985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992146694496306e-05,
+      "loss": 0.6667,
+      "step": 1986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921326703741976e-05,
+      "loss": 0.6468,
+      "step": 1987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921186337908466e-05,
+      "loss": 0.6539,
+      "step": 1988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921045847464305e-05,
+      "loss": 0.6343,
+      "step": 1989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992090523241125e-05,
+      "loss": 0.5798,
+      "step": 1990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992076449275107e-05,
+      "loss": 0.5464,
+      "step": 1991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920623628485535e-05,
+      "loss": 0.6167,
+      "step": 1992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992048263961641e-05,
+      "loss": 0.5085,
+      "step": 1993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920341526145467e-05,
+      "loss": 0.683,
+      "step": 1994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992020028807448e-05,
+      "loss": 0.6558,
+      "step": 1995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992005892540522e-05,
+      "loss": 0.591,
+      "step": 1996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919917438139466e-05,
+      "loss": 0.6321,
+      "step": 1997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991977582627899e-05,
+      "loss": 0.6297,
+      "step": 1998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991963408982558e-05,
+      "loss": 0.5792,
+      "step": 1999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919492228781007e-05,
+      "loss": 0.5969,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919350243147053e-05,
+      "loss": 0.6315,
+      "step": 2001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919208132925506e-05,
+      "loss": 0.6506,
+      "step": 2002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919065898118153e-05,
+      "loss": 0.6025,
+      "step": 2003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918923538726774e-05,
+      "loss": 0.5466,
+      "step": 2004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991878105475316e-05,
+      "loss": 0.5873,
+      "step": 2005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99186384461991e-05,
+      "loss": 0.5645,
+      "step": 2006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918495713066388e-05,
+      "loss": 0.6159,
+      "step": 2007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918352855356812e-05,
+      "loss": 0.612,
+      "step": 2008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918209873072166e-05,
+      "loss": 0.5775,
+      "step": 2009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918066766214256e-05,
+      "loss": 0.6579,
+      "step": 2010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917923534784867e-05,
+      "loss": 0.6471,
+      "step": 2011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917780178785802e-05,
+      "loss": 0.6011,
+      "step": 2012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917636698218866e-05,
+      "loss": 0.6296,
+      "step": 2013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917493093085856e-05,
+      "loss": 0.6042,
+      "step": 2014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917349363388577e-05,
+      "loss": 0.5903,
+      "step": 2015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917205509128835e-05,
+      "loss": 0.6183,
+      "step": 2016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917061530308435e-05,
+      "loss": 0.6638,
+      "step": 2017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991691742692919e-05,
+      "loss": 0.5618,
+      "step": 2018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99167731989929e-05,
+      "loss": 0.6157,
+      "step": 2019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916628846501388e-05,
+      "loss": 0.6096,
+      "step": 2020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916484369456464e-05,
+      "loss": 0.5404,
+      "step": 2021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991633976785994e-05,
+      "loss": 0.5791,
+      "step": 2022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916195041713632e-05,
+      "loss": 0.597,
+      "step": 2023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916050191019357e-05,
+      "loss": 0.5699,
+      "step": 2024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991590521577894e-05,
+      "loss": 0.618,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915760115994197e-05,
+      "loss": 0.6021,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915614891666946e-05,
+      "loss": 0.5956,
+      "step": 2027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915469542799026e-05,
+      "loss": 0.602,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915324069392247e-05,
+      "loss": 0.6153,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915178471448444e-05,
+      "loss": 0.7026,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915032748969445e-05,
+      "loss": 0.623,
+      "step": 2031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914886901957077e-05,
+      "loss": 0.6187,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914740930413177e-05,
+      "loss": 0.6274,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914594834339573e-05,
+      "loss": 0.6206,
+      "step": 2034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.6023,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914302268610606e-05,
+      "loss": 0.5836,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914155798958915e-05,
+      "loss": 0.6486,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914009204784873e-05,
+      "loss": 0.6907,
+      "step": 2038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991386248609032e-05,
+      "loss": 0.6239,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99137156428771e-05,
+      "loss": 0.5985,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913568675147053e-05,
+      "loss": 0.6708,
+      "step": 2041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913421582902032e-05,
+      "loss": 0.5624,
+      "step": 2042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913274366143876e-05,
+      "loss": 0.5608,
+      "step": 2043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913127024874442e-05,
+      "loss": 0.5943,
+      "step": 2044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991297955909558e-05,
+      "loss": 0.5435,
+      "step": 2045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991283196880913e-05,
+      "loss": 0.6002,
+      "step": 2046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912684254016964e-05,
+      "loss": 0.5985,
+      "step": 2047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991253641472093e-05,
+      "loss": 0.655,
+      "step": 2048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991238845092288e-05,
+      "loss": 0.6212,
+      "step": 2049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912240362624673e-05,
+      "loss": 0.6557,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912092149828175e-05,
+      "loss": 0.5658,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911943812535243e-05,
+      "loss": 0.7148,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911795350747742e-05,
+      "loss": 0.6642,
+      "step": 2053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911646764467537e-05,
+      "loss": 0.652,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911498053696493e-05,
+      "loss": 0.5936,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991134921843648e-05,
+      "loss": 0.6234,
+      "step": 2056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911200258689364e-05,
+      "loss": 0.5758,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991105117445702e-05,
+      "loss": 0.6338,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910901965741314e-05,
+      "loss": 0.6063,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910752632544125e-05,
+      "loss": 0.6313,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991060317486733e-05,
+      "loss": 0.5957,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910453592712802e-05,
+      "loss": 0.5571,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910303886082426e-05,
+      "loss": 0.648,
+      "step": 2063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910154054978075e-05,
+      "loss": 0.5962,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910004099401636e-05,
+      "loss": 0.621,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909854019354985e-05,
+      "loss": 0.6496,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.990970381484002e-05,
+      "loss": 0.6188,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909553485858615e-05,
+      "loss": 0.6359,
+      "step": 2068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909403032412665e-05,
+      "loss": 0.5967,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909252454504057e-05,
+      "loss": 0.579,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909101752134684e-05,
+      "loss": 0.6134,
+      "step": 2071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908950925306437e-05,
+      "loss": 0.6662,
+      "step": 2072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908799974021213e-05,
+      "loss": 0.6419,
+      "step": 2073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908648898280906e-05,
+      "loss": 0.6806,
+      "step": 2074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908497698087414e-05,
+      "loss": 0.598,
+      "step": 2075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908346373442637e-05,
+      "loss": 0.618,
+      "step": 2076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990819492434847e-05,
+      "loss": 0.5877,
+      "step": 2077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908043350806824e-05,
+      "loss": 0.5601,
+      "step": 2078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907891652819597e-05,
+      "loss": 0.6641,
+      "step": 2079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907739830388698e-05,
+      "loss": 0.6494,
+      "step": 2080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990758788351603e-05,
+      "loss": 0.5114,
+      "step": 2081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907435812203502e-05,
+      "loss": 0.6112,
+      "step": 2082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907283616453027e-05,
+      "loss": 0.6654,
+      "step": 2083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907131296266517e-05,
+      "loss": 0.6453,
+      "step": 2084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990697885164588e-05,
+      "loss": 0.6335,
+      "step": 2085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906826282593036e-05,
+      "loss": 0.593,
+      "step": 2086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906673589109897e-05,
+      "loss": 0.6309,
+      "step": 2087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990652077119838e-05,
+      "loss": 0.5943,
+      "step": 2088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906367828860414e-05,
+      "loss": 0.6005,
+      "step": 2089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990621476209791e-05,
+      "loss": 0.5896,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906061570912792e-05,
+      "loss": 0.6547,
+      "step": 2091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990590825530699e-05,
+      "loss": 0.5805,
+      "step": 2092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905754815282425e-05,
+      "loss": 0.6141,
+      "step": 2093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905601250841022e-05,
+      "loss": 0.6082,
+      "step": 2094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905447561984713e-05,
+      "loss": 0.5895,
+      "step": 2095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990529374871543e-05,
+      "loss": 0.6162,
+      "step": 2096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905139811035098e-05,
+      "loss": 0.6541,
+      "step": 2097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990498574894566e-05,
+      "loss": 0.6588,
+      "step": 2098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904831562449045e-05,
+      "loss": 0.6513,
+      "step": 2099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990467725154719e-05,
+      "loss": 0.7502,
+      "step": 2100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904522816242035e-05,
+      "loss": 0.6205,
+      "step": 2101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904368256535517e-05,
+      "loss": 0.5508,
+      "step": 2102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990421357242958e-05,
+      "loss": 0.6301,
+      "step": 2103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904058763926167e-05,
+      "loss": 0.6209,
+      "step": 2104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990390383102722e-05,
+      "loss": 0.6183,
+      "step": 2105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903748773734685e-05,
+      "loss": 0.6114,
+      "step": 2106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903593592050515e-05,
+      "loss": 0.6569,
+      "step": 2107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990343828597665e-05,
+      "loss": 0.5695,
+      "step": 2108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903282855515048e-05,
+      "loss": 0.6195,
+      "step": 2109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903127300667656e-05,
+      "loss": 0.5554,
+      "step": 2110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902971621436435e-05,
+      "loss": 0.5738,
+      "step": 2111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902815817823334e-05,
+      "loss": 0.5798,
+      "step": 2112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902659889830313e-05,
+      "loss": 0.6505,
+      "step": 2113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902503837459323e-05,
+      "loss": 0.6426,
+      "step": 2114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902347660712336e-05,
+      "loss": 0.6378,
+      "step": 2115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902191359591306e-05,
+      "loss": 0.637,
+      "step": 2116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99020349340982e-05,
+      "loss": 0.5622,
+      "step": 2117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901878384234975e-05,
+      "loss": 0.579,
+      "step": 2118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990172171000361e-05,
+      "loss": 0.5432,
+      "step": 2119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901564911406063e-05,
+      "loss": 0.6536,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901407988444306e-05,
+      "loss": 0.5267,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990125094112031e-05,
+      "loss": 0.7341,
+      "step": 2122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901093769436045e-05,
+      "loss": 0.5916,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990093647339349e-05,
+      "loss": 0.5978,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990077905299462e-05,
+      "loss": 0.5524,
+      "step": 2125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990062150824141e-05,
+      "loss": 0.6532,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900463839135838e-05,
+      "loss": 0.6148,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900306045679887e-05,
+      "loss": 0.5796,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900148127875535e-05,
+      "loss": 0.6255,
+      "step": 2129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989999008572477e-05,
+      "loss": 0.6236,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899831919229574e-05,
+      "loss": 0.5885,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989967362839194e-05,
+      "loss": 0.6224,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899515213213845e-05,
+      "loss": 0.6073,
+      "step": 2133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899356673697285e-05,
+      "loss": 0.619,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899198009844253e-05,
+      "loss": 0.5407,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989903922165674e-05,
+      "loss": 0.6245,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898880309136735e-05,
+      "loss": 0.6583,
+      "step": 2137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898721272286243e-05,
+      "loss": 0.6215,
+      "step": 2138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898562111107255e-05,
+      "loss": 0.655,
+      "step": 2139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898402825601775e-05,
+      "loss": 0.5651,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98982434157718e-05,
+      "loss": 0.6127,
+      "step": 2141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898083881619334e-05,
+      "loss": 0.6345,
+      "step": 2142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897924223146378e-05,
+      "loss": 0.7236,
+      "step": 2143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897764440354942e-05,
+      "loss": 0.5642,
+      "step": 2144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989760453324703e-05,
+      "loss": 0.6027,
+      "step": 2145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897444501824647e-05,
+      "loss": 0.6441,
+      "step": 2146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989728434608981e-05,
+      "loss": 0.6016,
+      "step": 2147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989712406604453e-05,
+      "loss": 0.5838,
+      "step": 2148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896963661690813e-05,
+      "loss": 0.5899,
+      "step": 2149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989680313303068e-05,
+      "loss": 0.5789,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896642480066145e-05,
+      "loss": 0.59,
+      "step": 2151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896481702799225e-05,
+      "loss": 0.6327,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896320801231944e-05,
+      "loss": 0.4911,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896159775366315e-05,
+      "loss": 0.6155,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895998625204368e-05,
+      "loss": 0.6285,
+      "step": 2155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895837350748125e-05,
+      "loss": 0.5614,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895675951999606e-05,
+      "loss": 0.6039,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895514428960847e-05,
+      "loss": 0.659,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895352781633874e-05,
+      "loss": 0.6193,
+      "step": 2159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989519101002071e-05,
+      "loss": 0.6339,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895029114123396e-05,
+      "loss": 0.5994,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989486709394396e-05,
+      "loss": 0.6485,
+      "step": 2162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894704949484444e-05,
+      "loss": 0.6191,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894542680746877e-05,
+      "loss": 0.6451,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98943802877333e-05,
+      "loss": 0.677,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894217770445753e-05,
+      "loss": 0.7001,
+      "step": 2166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894055128886275e-05,
+      "loss": 0.6027,
+      "step": 2167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989389236305691e-05,
+      "loss": 0.631,
+      "step": 2168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893729472959703e-05,
+      "loss": 0.5923,
+      "step": 2169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98935664585967e-05,
+      "loss": 0.552,
+      "step": 2170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893403319969948e-05,
+      "loss": 0.6111,
+      "step": 2171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98932400570815e-05,
+      "loss": 0.5663,
+      "step": 2172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98930766699334e-05,
+      "loss": 0.5272,
+      "step": 2173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98929131585277e-05,
+      "loss": 0.5993,
+      "step": 2174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989274952286646e-05,
+      "loss": 0.5791,
+      "step": 2175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989258576295173e-05,
+      "loss": 0.6099,
+      "step": 2176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892421878785573e-05,
+      "loss": 0.5977,
+      "step": 2177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892257870370038e-05,
+      "loss": 0.5658,
+      "step": 2178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989209373770719e-05,
+      "loss": 0.614,
+      "step": 2179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891929480799097e-05,
+      "loss": 0.6894,
+      "step": 2180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989176509964781e-05,
+      "loss": 0.6617,
+      "step": 2181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891600594255403e-05,
+      "loss": 0.6379,
+      "step": 2182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989143596462394e-05,
+      "loss": 0.5958,
+      "step": 2183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989127121075548e-05,
+      "loss": 0.6927,
+      "step": 2184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891106332652105e-05,
+      "loss": 0.5709,
+      "step": 2185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890941330315878e-05,
+      "loss": 0.665,
+      "step": 2186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890776203748877e-05,
+      "loss": 0.6683,
+      "step": 2187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989061095295317e-05,
+      "loss": 0.6172,
+      "step": 2188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890445577930836e-05,
+      "loss": 0.5132,
+      "step": 2189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890280078683954e-05,
+      "loss": 0.505,
+      "step": 2190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890114455214594e-05,
+      "loss": 0.61,
+      "step": 2191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889948707524848e-05,
+      "loss": 0.6637,
+      "step": 2192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988978283561679e-05,
+      "loss": 0.5488,
+      "step": 2193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889616839492508e-05,
+      "loss": 0.625,
+      "step": 2194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889450719154086e-05,
+      "loss": 0.6677,
+      "step": 2195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889284474603605e-05,
+      "loss": 0.6264,
+      "step": 2196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988911810584316e-05,
+      "loss": 0.6237,
+      "step": 2197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888951612874834e-05,
+      "loss": 0.583,
+      "step": 2198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888784995700726e-05,
+      "loss": 0.6196,
+      "step": 2199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888618254322924e-05,
+      "loss": 0.6707,
+      "step": 2200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888451388743522e-05,
+      "loss": 0.6243,
+      "step": 2201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888284398964614e-05,
+      "loss": 0.6142,
+      "step": 2202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888117284988304e-05,
+      "loss": 0.5885,
+      "step": 2203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887950046816687e-05,
+      "loss": 0.6225,
+      "step": 2204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988778268445186e-05,
+      "loss": 0.6228,
+      "step": 2205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988761519789593e-05,
+      "loss": 0.5771,
+      "step": 2206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887447587151002e-05,
+      "loss": 0.584,
+      "step": 2207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988727985221918e-05,
+      "loss": 0.6342,
+      "step": 2208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887111993102564e-05,
+      "loss": 0.6508,
+      "step": 2209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988694400980327e-05,
+      "loss": 0.5832,
+      "step": 2210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886775902323405e-05,
+      "loss": 0.5692,
+      "step": 2211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988660767066508e-05,
+      "loss": 0.5707,
+      "step": 2212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886439314830413e-05,
+      "loss": 0.6817,
+      "step": 2213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988627083482151e-05,
+      "loss": 0.5618,
+      "step": 2214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886102230640497e-05,
+      "loss": 0.5795,
+      "step": 2215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988593350228948e-05,
+      "loss": 0.5331,
+      "step": 2216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988576464977059e-05,
+      "loss": 0.6119,
+      "step": 2217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885595673085937e-05,
+      "loss": 0.7157,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988542657223765e-05,
+      "loss": 0.5805,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885257347227855e-05,
+      "loss": 0.6765,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885087998058673e-05,
+      "loss": 0.648,
+      "step": 2221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988491852473223e-05,
+      "loss": 0.5822,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884748927250658e-05,
+      "loss": 0.6229,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884579205616088e-05,
+      "loss": 0.6098,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884409359830648e-05,
+      "loss": 0.6361,
+      "step": 2225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988423938989647e-05,
+      "loss": 0.5878,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884069295815696e-05,
+      "loss": 0.6524,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883899077590453e-05,
+      "loss": 0.6122,
+      "step": 2228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883728735222888e-05,
+      "loss": 0.567,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883558268715137e-05,
+      "loss": 0.6212,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988338767806934e-05,
+      "loss": 0.64,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883216963287637e-05,
+      "loss": 0.5664,
+      "step": 2232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883046124372177e-05,
+      "loss": 0.6226,
+      "step": 2233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882875161325106e-05,
+      "loss": 0.5825,
+      "step": 2234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988270407414857e-05,
+      "loss": 0.5894,
+      "step": 2235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882532862844714e-05,
+      "loss": 0.5513,
+      "step": 2236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882361527415694e-05,
+      "loss": 0.6092,
+      "step": 2237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882190067863656e-05,
+      "loss": 0.5656,
+      "step": 2238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882018484190765e-05,
+      "loss": 0.6059,
+      "step": 2239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881846776399164e-05,
+      "loss": 0.6151,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881674944491012e-05,
+      "loss": 0.6514,
+      "step": 2241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988150298846847e-05,
+      "loss": 0.5681,
+      "step": 2242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881330908333702e-05,
+      "loss": 0.554,
+      "step": 2243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988115870408886e-05,
+      "loss": 0.5697,
+      "step": 2244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988098637573611e-05,
+      "loss": 0.5929,
+      "step": 2245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880813923277622e-05,
+      "loss": 0.5491,
+      "step": 2246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880641346715555e-05,
+      "loss": 0.562,
+      "step": 2247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988046864605208e-05,
+      "loss": 0.62,
+      "step": 2248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880295821289363e-05,
+      "loss": 0.6477,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988012287242958e-05,
+      "loss": 0.5765,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98799497994749e-05,
+      "loss": 0.5432,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879776602427498e-05,
+      "loss": 0.5912,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879603281289545e-05,
+      "loss": 0.5973,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879429836063228e-05,
+      "loss": 0.6194,
+      "step": 2254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879256266750713e-05,
+      "loss": 0.6189,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879082573354188e-05,
+      "loss": 0.6116,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878908755875833e-05,
+      "loss": 0.626,
+      "step": 2257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878734814317827e-05,
+      "loss": 0.5554,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878560748682363e-05,
+      "loss": 0.5748,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878386558971622e-05,
+      "loss": 0.5955,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987821224518779e-05,
+      "loss": 0.6332,
+      "step": 2261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987803780733306e-05,
+      "loss": 0.5985,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9877863245409626e-05,
+      "loss": 0.5838,
+      "step": 2263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877688559419673e-05,
+      "loss": 0.5655,
+      "step": 2264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98775137493654e-05,
+      "loss": 0.5582,
+      "step": 2265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877338815249e-05,
+      "loss": 0.5639,
+      "step": 2266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877163757072667e-05,
+      "loss": 0.6719,
+      "step": 2267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876988574838608e-05,
+      "loss": 0.6961,
+      "step": 2268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876813268549022e-05,
+      "loss": 0.608,
+      "step": 2269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98766378382061e-05,
+      "loss": 0.5438,
+      "step": 2270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876462283812065e-05,
+      "loss": 0.6037,
+      "step": 2271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98762866053691e-05,
+      "loss": 0.5769,
+      "step": 2272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987611080287943e-05,
+      "loss": 0.6214,
+      "step": 2273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875934876345254e-05,
+      "loss": 0.575,
+      "step": 2274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987575882576878e-05,
+      "loss": 0.6322,
+      "step": 2275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875582651152222e-05,
+      "loss": 0.611,
+      "step": 2276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987540635249779e-05,
+      "loss": 0.5218,
+      "step": 2277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987522992980771e-05,
+      "loss": 0.6413,
+      "step": 2278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875053383084183e-05,
+      "loss": 0.6128,
+      "step": 2279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874876712329433e-05,
+      "loss": 0.6037,
+      "step": 2280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874699917545677e-05,
+      "loss": 0.5668,
+      "step": 2281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987452299873514e-05,
+      "loss": 0.5503,
+      "step": 2282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874345955900042e-05,
+      "loss": 0.5846,
+      "step": 2283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98741687890426e-05,
+      "loss": 0.597,
+      "step": 2284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873991498165052e-05,
+      "loss": 0.6388,
+      "step": 2285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873814083269615e-05,
+      "loss": 0.6263,
+      "step": 2286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873636544358518e-05,
+      "loss": 0.5941,
+      "step": 2287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873458881433995e-05,
+      "loss": 0.5839,
+      "step": 2288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873281094498275e-05,
+      "loss": 0.5729,
+      "step": 2289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987310318355359e-05,
+      "loss": 0.638,
+      "step": 2290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987292514860218e-05,
+      "loss": 0.5739,
+      "step": 2291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987274698964627e-05,
+      "loss": 0.5772,
+      "step": 2292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872568706688115e-05,
+      "loss": 0.634,
+      "step": 2293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872390299729936e-05,
+      "loss": 0.628,
+      "step": 2294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872211768773987e-05,
+      "loss": 0.6046,
+      "step": 2295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98720331138225e-05,
+      "loss": 0.5913,
+      "step": 2296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871854334877728e-05,
+      "loss": 0.628,
+      "step": 2297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987167543194191e-05,
+      "loss": 0.6039,
+      "step": 2298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871496405017298e-05,
+      "loss": 0.6058,
+      "step": 2299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871317254106137e-05,
+      "loss": 0.5541,
+      "step": 2300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871137979210678e-05,
+      "loss": 0.5629,
+      "step": 2301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870958580333177e-05,
+      "loss": 0.5723,
+      "step": 2302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987077905747588e-05,
+      "loss": 0.741,
+      "step": 2303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870599410641047e-05,
+      "loss": 0.6191,
+      "step": 2304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870419639830932e-05,
+      "loss": 0.6228,
+      "step": 2305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870239745047793e-05,
+      "loss": 0.5938,
+      "step": 2306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870059726293892e-05,
+      "loss": 0.6818,
+      "step": 2307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869879583571484e-05,
+      "loss": 0.6469,
+      "step": 2308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869699316882836e-05,
+      "loss": 0.612,
+      "step": 2309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869518926230212e-05,
+      "loss": 0.5819,
+      "step": 2310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986933841161588e-05,
+      "loss": 0.5986,
+      "step": 2311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869157773042107e-05,
+      "loss": 0.5935,
+      "step": 2312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868977010511153e-05,
+      "loss": 0.5868,
+      "step": 2313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868796124025303e-05,
+      "loss": 0.6255,
+      "step": 2314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868615113586816e-05,
+      "loss": 0.5674,
+      "step": 2315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986843397919797e-05,
+      "loss": 0.5788,
+      "step": 2316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986825272086104e-05,
+      "loss": 0.5899,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868071338578305e-05,
+      "loss": 0.6085,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867889832352043e-05,
+      "loss": 0.6292,
+      "step": 2319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867708202184526e-05,
+      "loss": 0.5281,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867526448078047e-05,
+      "loss": 0.6326,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867344570034883e-05,
+      "loss": 0.6366,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867162568057318e-05,
+      "loss": 0.6533,
+      "step": 2323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866980442147638e-05,
+      "loss": 0.5887,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986679819230813e-05,
+      "loss": 0.6372,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866615818541085e-05,
+      "loss": 0.6223,
+      "step": 2326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866433320848793e-05,
+      "loss": 0.5712,
+      "step": 2327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866250699233545e-05,
+      "loss": 0.6065,
+      "step": 2328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866067953697637e-05,
+      "loss": 0.5922,
+      "step": 2329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986588508424336e-05,
+      "loss": 0.6311,
+      "step": 2330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865702090873016e-05,
+      "loss": 0.6351,
+      "step": 2331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98655189735889e-05,
+      "loss": 0.62,
+      "step": 2332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865335732393313e-05,
+      "loss": 0.6305,
+      "step": 2333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986515236728856e-05,
+      "loss": 0.6751,
+      "step": 2334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986496887827694e-05,
+      "loss": 0.6752,
+      "step": 2335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864785265360754e-05,
+      "loss": 0.5859,
+      "step": 2336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864601528542317e-05,
+      "loss": 0.6864,
+      "step": 2337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986441766782393e-05,
+      "loss": 0.5351,
+      "step": 2338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864233683207907e-05,
+      "loss": 0.6128,
+      "step": 2339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864049574696556e-05,
+      "loss": 0.6021,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986386534229219e-05,
+      "loss": 0.6967,
+      "step": 2341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863680985997118e-05,
+      "loss": 0.5754,
+      "step": 2342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863496505813666e-05,
+      "loss": 0.6261,
+      "step": 2343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863311901744144e-05,
+      "loss": 0.6331,
+      "step": 2344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986312717379087e-05,
+      "loss": 0.6334,
+      "step": 2345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862942321956167e-05,
+      "loss": 0.5806,
+      "step": 2346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862757346242357e-05,
+      "loss": 0.6197,
+      "step": 2347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862572246651764e-05,
+      "loss": 0.5849,
+      "step": 2348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986238702318671e-05,
+      "loss": 0.6553,
+      "step": 2349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986220167584952e-05,
+      "loss": 0.6111,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862016204642528e-05,
+      "loss": 0.6053,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986183060956806e-05,
+      "loss": 0.5746,
+      "step": 2352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861644890628446e-05,
+      "loss": 0.5943,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986145904782602e-05,
+      "loss": 0.6006,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861273081163116e-05,
+      "loss": 0.6201,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861086990642068e-05,
+      "loss": 0.6289,
+      "step": 2356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860900776265216e-05,
+      "loss": 0.5772,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98607144380349e-05,
+      "loss": 0.6027,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986052797595346e-05,
+      "loss": 0.6159,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986034139002323e-05,
+      "loss": 0.6131,
+      "step": 2360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860154680246567e-05,
+      "loss": 0.5792,
+      "step": 2361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859967846625805e-05,
+      "loss": 0.5642,
+      "step": 2362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859780889163297e-05,
+      "loss": 0.6186,
+      "step": 2363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859593807861386e-05,
+      "loss": 0.5965,
+      "step": 2364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859406602722426e-05,
+      "loss": 0.5834,
+      "step": 2365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859219273748766e-05,
+      "loss": 0.5155,
+      "step": 2366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985903182094276e-05,
+      "loss": 0.6123,
+      "step": 2367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858844244306763e-05,
+      "loss": 0.6013,
+      "step": 2368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985865654384313e-05,
+      "loss": 0.584,
+      "step": 2369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858468719554216e-05,
+      "loss": 0.6014,
+      "step": 2370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858280771442387e-05,
+      "loss": 0.6377,
+      "step": 2371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858092699509996e-05,
+      "loss": 0.5993,
+      "step": 2372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985790450375941e-05,
+      "loss": 0.5946,
+      "step": 2373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985771618419299e-05,
+      "loss": 0.6516,
+      "step": 2374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857527740813106e-05,
+      "loss": 0.6245,
+      "step": 2375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985733917362212e-05,
+      "loss": 0.6541,
+      "step": 2376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98571504826224e-05,
+      "loss": 0.6533,
+      "step": 2377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985696166781632e-05,
+      "loss": 0.5461,
+      "step": 2378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985677272920625e-05,
+      "loss": 0.6199,
+      "step": 2379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985658366679456e-05,
+      "loss": 0.6067,
+      "step": 2380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985639448058363e-05,
+      "loss": 0.6072,
+      "step": 2381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856205170575832e-05,
+      "loss": 0.557,
+      "step": 2382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856015736773542e-05,
+      "loss": 0.6699,
+      "step": 2383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855826179179146e-05,
+      "loss": 0.6485,
+      "step": 2384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855636497795017e-05,
+      "loss": 0.6694,
+      "step": 2385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985544669262355e-05,
+      "loss": 0.5857,
+      "step": 2386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855256763667113e-05,
+      "loss": 0.6865,
+      "step": 2387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98550667109281e-05,
+      "loss": 0.5701,
+      "step": 2388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98548765344089e-05,
+      "loss": 0.6351,
+      "step": 2389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854686234111897e-05,
+      "loss": 0.5713,
+      "step": 2390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854495810039482e-05,
+      "loss": 0.6176,
+      "step": 2391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854305262194052e-05,
+      "loss": 0.6347,
+      "step": 2392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985411459057799e-05,
+      "loss": 0.6836,
+      "step": 2393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98539237951937e-05,
+      "loss": 0.5944,
+      "step": 2394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853732876043575e-05,
+      "loss": 0.5417,
+      "step": 2395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853541833130014e-05,
+      "loss": 0.5542,
+      "step": 2396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853350666455418e-05,
+      "loss": 0.5923,
+      "step": 2397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985315937602218e-05,
+      "loss": 0.6369,
+      "step": 2398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852967961832713e-05,
+      "loss": 0.5875,
+      "step": 2399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852776423889414e-05,
+      "loss": 0.5865,
+      "step": 2400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852584762194693e-05,
+      "loss": 0.6336,
+      "step": 2401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852392976750952e-05,
+      "loss": 0.5994,
+      "step": 2402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852201067560607e-05,
+      "loss": 0.5997,
+      "step": 2403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852009034626067e-05,
+      "loss": 0.5776,
+      "step": 2404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851816877949737e-05,
+      "loss": 0.6529,
+      "step": 2405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985162459753404e-05,
+      "loss": 0.5637,
+      "step": 2406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851432193381378e-05,
+      "loss": 0.5294,
+      "step": 2407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851239665494184e-05,
+      "loss": 0.5693,
+      "step": 2408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851047013874866e-05,
+      "loss": 0.5169,
+      "step": 2409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850854238525844e-05,
+      "loss": 0.5856,
+      "step": 2410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985066133944954e-05,
+      "loss": 0.6188,
+      "step": 2411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985046831664838e-05,
+      "loss": 0.6117,
+      "step": 2412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985027517012479e-05,
+      "loss": 0.5197,
+      "step": 2413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850081899881186e-05,
+      "loss": 0.6148,
+      "step": 2414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849888505920004e-05,
+      "loss": 0.6626,
+      "step": 2415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984969498824367e-05,
+      "loss": 0.6485,
+      "step": 2416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849501346854615e-05,
+      "loss": 0.5907,
+      "step": 2417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984930758175527e-05,
+      "loss": 0.6067,
+      "step": 2418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849113692948078e-05,
+      "loss": 0.585,
+      "step": 2419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848919680435462e-05,
+      "loss": 0.6095,
+      "step": 2420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984872554421986e-05,
+      "loss": 0.662,
+      "step": 2421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848531284303715e-05,
+      "loss": 0.6308,
+      "step": 2422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848336900689465e-05,
+      "loss": 0.6257,
+      "step": 2423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848142393379555e-05,
+      "loss": 0.6174,
+      "step": 2424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847947762376424e-05,
+      "loss": 0.6129,
+      "step": 2425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984775300768252e-05,
+      "loss": 0.5913,
+      "step": 2426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847558129300282e-05,
+      "loss": 0.6476,
+      "step": 2427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847363127232163e-05,
+      "loss": 0.6765,
+      "step": 2428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984716800148061e-05,
+      "loss": 0.5947,
+      "step": 2429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984697275204808e-05,
+      "loss": 0.6984,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984677737893702e-05,
+      "loss": 0.6278,
+      "step": 2431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984658188214988e-05,
+      "loss": 0.578,
+      "step": 2432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846386261689123e-05,
+      "loss": 0.6852,
+      "step": 2433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846190517557208e-05,
+      "loss": 0.6118,
+      "step": 2434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984599464975658e-05,
+      "loss": 0.5485,
+      "step": 2435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845798658289714e-05,
+      "loss": 0.5759,
+      "step": 2436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845602543159065e-05,
+      "loss": 0.6645,
+      "step": 2437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984540630436709e-05,
+      "loss": 0.5846,
+      "step": 2438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984520994191627e-05,
+      "loss": 0.5719,
+      "step": 2439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845013455809055e-05,
+      "loss": 0.632,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844816846047923e-05,
+      "loss": 0.6137,
+      "step": 2441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984462011263534e-05,
+      "loss": 0.5987,
+      "step": 2442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844423255573773e-05,
+      "loss": 0.5978,
+      "step": 2443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844226274865705e-05,
+      "loss": 0.6589,
+      "step": 2444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844029170513598e-05,
+      "loss": 0.6166,
+      "step": 2445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843831942519936e-05,
+      "loss": 0.6909,
+      "step": 2446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984363459088719e-05,
+      "loss": 0.6006,
+      "step": 2447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843437115617848e-05,
+      "loss": 0.639,
+      "step": 2448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984323951671438e-05,
+      "loss": 0.604,
+      "step": 2449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843041794179272e-05,
+      "loss": 0.5192,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842843948015013e-05,
+      "loss": 0.6369,
+      "step": 2451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984264597822408e-05,
+      "loss": 0.6909,
+      "step": 2452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984244788480896e-05,
+      "loss": 0.6418,
+      "step": 2453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842249667772144e-05,
+      "loss": 0.563,
+      "step": 2454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842051327116122e-05,
+      "loss": 0.6028,
+      "step": 2455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9841852862843386e-05,
+      "loss": 0.5297,
+      "step": 2456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841654274956423e-05,
+      "loss": 0.5959,
+      "step": 2457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841455563457734e-05,
+      "loss": 0.6254,
+      "step": 2458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984125672834981e-05,
+      "loss": 0.6364,
+      "step": 2459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984105776963515e-05,
+      "loss": 0.6055,
+      "step": 2460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984085868731626e-05,
+      "loss": 0.5537,
+      "step": 2461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840659481395623e-05,
+      "loss": 0.6059,
+      "step": 2462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984046015187576e-05,
+      "loss": 0.5967,
+      "step": 2463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840260698759163e-05,
+      "loss": 0.6219,
+      "step": 2464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984006112204834e-05,
+      "loss": 0.5716,
+      "step": 2465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98398614217458e-05,
+      "loss": 0.6341,
+      "step": 2466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839661597854048e-05,
+      "loss": 0.587,
+      "step": 2467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98394616503756e-05,
+      "loss": 0.6627,
+      "step": 2468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839261579312956e-05,
+      "loss": 0.6367,
+      "step": 2469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839061384668637e-05,
+      "loss": 0.5599,
+      "step": 2470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838861066445158e-05,
+      "loss": 0.5371,
+      "step": 2471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983866062464503e-05,
+      "loss": 0.5823,
+      "step": 2472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838460059270775e-05,
+      "loss": 0.5775,
+      "step": 2473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838259370324913e-05,
+      "loss": 0.6033,
+      "step": 2474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838058557809957e-05,
+      "loss": 0.5216,
+      "step": 2475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983785762172844e-05,
+      "loss": 0.5768,
+      "step": 2476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983765656208288e-05,
+      "loss": 0.5946,
+      "step": 2477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98374553788758e-05,
+      "loss": 0.6416,
+      "step": 2478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983725407210973e-05,
+      "loss": 0.5879,
+      "step": 2479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837052641787197e-05,
+      "loss": 0.5604,
+      "step": 2480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836851087910733e-05,
+      "loss": 0.6579,
+      "step": 2481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983664941048287e-05,
+      "loss": 0.5832,
+      "step": 2482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983644760950614e-05,
+      "loss": 0.5714,
+      "step": 2483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983624568498307e-05,
+      "loss": 0.579,
+      "step": 2484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983604363691621e-05,
+      "loss": 0.6104,
+      "step": 2485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983584146530809e-05,
+      "loss": 0.5507,
+      "step": 2486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983563917016125e-05,
+      "loss": 0.5071,
+      "step": 2487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983543675147823e-05,
+      "loss": 0.6355,
+      "step": 2488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835234209261576e-05,
+      "loss": 0.6228,
+      "step": 2489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835031543513826e-05,
+      "loss": 0.6723,
+      "step": 2490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834828754237532e-05,
+      "loss": 0.591,
+      "step": 2491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983462584143524e-05,
+      "loss": 0.5953,
+      "step": 2492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983442280510949e-05,
+      "loss": 0.6406,
+      "step": 2493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983421964526284e-05,
+      "loss": 0.6132,
+      "step": 2494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834016361897846e-05,
+      "loss": 0.6618,
+      "step": 2495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833812955017054e-05,
+      "loss": 0.6077,
+      "step": 2496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833609424623018e-05,
+      "loss": 0.5806,
+      "step": 2497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833405770718297e-05,
+      "loss": 0.6645,
+      "step": 2498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833201993305452e-05,
+      "loss": 0.6513,
+      "step": 2499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832998092387035e-05,
+      "loss": 0.5565,
+      "step": 2500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832794067965617e-05,
+      "loss": 0.5584,
+      "step": 2501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983258992004375e-05,
+      "loss": 0.5637,
+      "step": 2502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832385648624003e-05,
+      "loss": 0.6022,
+      "step": 2503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983218125370894e-05,
+      "loss": 0.5568,
+      "step": 2504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831976735301136e-05,
+      "loss": 0.6352,
+      "step": 2505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983177209340315e-05,
+      "loss": 0.6125,
+      "step": 2506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831567328017556e-05,
+      "loss": 0.6332,
+      "step": 2507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831362439146922e-05,
+      "loss": 0.58,
+      "step": 2508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983115742679383e-05,
+      "loss": 0.599,
+      "step": 2509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983095229096085e-05,
+      "loss": 0.6061,
+      "step": 2510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830747031650557e-05,
+      "loss": 0.5483,
+      "step": 2511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830541648865532e-05,
+      "loss": 0.6526,
+      "step": 2512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983033614260835e-05,
+      "loss": 0.6192,
+      "step": 2513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830130512881598e-05,
+      "loss": 0.5899,
+      "step": 2514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982992475968786e-05,
+      "loss": 0.5889,
+      "step": 2515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982971888302971e-05,
+      "loss": 0.7006,
+      "step": 2516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829512882909743e-05,
+      "loss": 0.6231,
+      "step": 2517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829306759330542e-05,
+      "loss": 0.6245,
+      "step": 2518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98291005122947e-05,
+      "loss": 0.5753,
+      "step": 2519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828894141804804e-05,
+      "loss": 0.6361,
+      "step": 2520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982868764786345e-05,
+      "loss": 0.5999,
+      "step": 2521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828481030473226e-05,
+      "loss": 0.5629,
+      "step": 2522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982827428963673e-05,
+      "loss": 0.7043,
+      "step": 2523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982806742535656e-05,
+      "loss": 0.6006,
+      "step": 2524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827860437635308e-05,
+      "loss": 0.5812,
+      "step": 2525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827653326475583e-05,
+      "loss": 0.5771,
+      "step": 2526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982744609187998e-05,
+      "loss": 0.5197,
+      "step": 2527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827238733851103e-05,
+      "loss": 0.606,
+      "step": 2528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982703125239156e-05,
+      "loss": 0.5943,
+      "step": 2529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982682364750395e-05,
+      "loss": 0.6077,
+      "step": 2530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826615919190886e-05,
+      "loss": 0.5963,
+      "step": 2531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826408067454975e-05,
+      "loss": 0.6001,
+      "step": 2532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826200092298832e-05,
+      "loss": 0.5716,
+      "step": 2533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825991993725066e-05,
+      "loss": 0.6286,
+      "step": 2534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825783771736288e-05,
+      "loss": 0.6756,
+      "step": 2535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982557542633511e-05,
+      "loss": 0.645,
+      "step": 2536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825366957524162e-05,
+      "loss": 0.6001,
+      "step": 2537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982515836530605e-05,
+      "loss": 0.6417,
+      "step": 2538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824949649683406e-05,
+      "loss": 0.6143,
+      "step": 2539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824740810658838e-05,
+      "loss": 0.6264,
+      "step": 2540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824531848234976e-05,
+      "loss": 0.5642,
+      "step": 2541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824322762414443e-05,
+      "loss": 0.6242,
+      "step": 2542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982411355319987e-05,
+      "loss": 0.5987,
+      "step": 2543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823904220593874e-05,
+      "loss": 0.6451,
+      "step": 2544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823694764599096e-05,
+      "loss": 0.6415,
+      "step": 2545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982348518521816e-05,
+      "loss": 0.5423,
+      "step": 2546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98232754824537e-05,
+      "loss": 0.5564,
+      "step": 2547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823065656308345e-05,
+      "loss": 0.5613,
+      "step": 2548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822855706784742e-05,
+      "loss": 0.5572,
+      "step": 2549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822645633885517e-05,
+      "loss": 0.616,
+      "step": 2550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822435437613315e-05,
+      "loss": 0.5971,
+      "step": 2551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822225117970774e-05,
+      "loss": 0.7041,
+      "step": 2552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822014674960534e-05,
+      "loss": 0.666,
+      "step": 2553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821804108585244e-05,
+      "loss": 0.5825,
+      "step": 2554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821593418847537e-05,
+      "loss": 0.6035,
+      "step": 2555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821382605750073e-05,
+      "loss": 0.578,
+      "step": 2556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982117166929549e-05,
+      "loss": 0.5834,
+      "step": 2557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820960609486443e-05,
+      "loss": 0.5678,
+      "step": 2558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982074942632558e-05,
+      "loss": 0.5556,
+      "step": 2559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982053811981555e-05,
+      "loss": 0.545,
+      "step": 2560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820326689959014e-05,
+      "loss": 0.5577,
+      "step": 2561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820115136758627e-05,
+      "loss": 0.5522,
+      "step": 2562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981990346021704e-05,
+      "loss": 0.5689,
+      "step": 2563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981969166033692e-05,
+      "loss": 0.6117,
+      "step": 2564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981947973712092e-05,
+      "loss": 0.5768,
+      "step": 2565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819267690571706e-05,
+      "loss": 0.6252,
+      "step": 2566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981905552069194e-05,
+      "loss": 0.5741,
+      "step": 2567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818843227484283e-05,
+      "loss": 0.6283,
+      "step": 2568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981863081095141e-05,
+      "loss": 0.6537,
+      "step": 2569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981841827109598e-05,
+      "loss": 0.5874,
+      "step": 2570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818205607920672e-05,
+      "loss": 0.6366,
+      "step": 2571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817992821428148e-05,
+      "loss": 0.5842,
+      "step": 2572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981777991162108e-05,
+      "loss": 0.5467,
+      "step": 2573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817566878502154e-05,
+      "loss": 0.627,
+      "step": 2574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817353722074035e-05,
+      "loss": 0.6003,
+      "step": 2575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817140442339406e-05,
+      "loss": 0.591,
+      "step": 2576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816927039300937e-05,
+      "loss": 0.6284,
+      "step": 2577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816713512961318e-05,
+      "loss": 0.6246,
+      "step": 2578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981649986332323e-05,
+      "loss": 0.5321,
+      "step": 2579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816286090389353e-05,
+      "loss": 0.5795,
+      "step": 2580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981607219416237e-05,
+      "loss": 0.615,
+      "step": 2581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981585817464497e-05,
+      "loss": 0.5653,
+      "step": 2582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815644031839847e-05,
+      "loss": 0.5948,
+      "step": 2583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981542976574968e-05,
+      "loss": 0.6206,
+      "step": 2584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981521537637717e-05,
+      "loss": 0.5592,
+      "step": 2585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815000863725e-05,
+      "loss": 0.6617,
+      "step": 2586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981478622779587e-05,
+      "loss": 0.5736,
+      "step": 2587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981457146859248e-05,
+      "loss": 0.5977,
+      "step": 2588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981435658611752e-05,
+      "loss": 0.6726,
+      "step": 2589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981414158037369e-05,
+      "loss": 0.556,
+      "step": 2590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813926451363693e-05,
+      "loss": 0.6365,
+      "step": 2591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813711199090232e-05,
+      "loss": 0.6865,
+      "step": 2592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813495823556012e-05,
+      "loss": 0.6174,
+      "step": 2593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981328032476373e-05,
+      "loss": 0.5991,
+      "step": 2594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813064702716094e-05,
+      "loss": 0.612,
+      "step": 2595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981284895741582e-05,
+      "loss": 0.7168,
+      "step": 2596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812633088865612e-05,
+      "loss": 0.6404,
+      "step": 2597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812417097068187e-05,
+      "loss": 0.5594,
+      "step": 2598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812200982026248e-05,
+      "loss": 0.6087,
+      "step": 2599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981198474374252e-05,
+      "loss": 0.5826,
+      "step": 2600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811768382219712e-05,
+      "loss": 0.6168,
+      "step": 2601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811551897460543e-05,
+      "loss": 0.5818,
+      "step": 2602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811335289467733e-05,
+      "loss": 0.5524,
+      "step": 2603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811118558244003e-05,
+      "loss": 0.6035,
+      "step": 2604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810901703792073e-05,
+      "loss": 0.6428,
+      "step": 2605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981068472611467e-05,
+      "loss": 0.6023,
+      "step": 2606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810467625214516e-05,
+      "loss": 0.5832,
+      "step": 2607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810250401094338e-05,
+      "loss": 0.6359,
+      "step": 2608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981003305375687e-05,
+      "loss": 0.5947,
+      "step": 2609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809815583204833e-05,
+      "loss": 0.5943,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809597989440964e-05,
+      "loss": 0.6064,
+      "step": 2611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809380272467994e-05,
+      "loss": 0.6003,
+      "step": 2612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980916243228866e-05,
+      "loss": 0.5878,
+      "step": 2613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808944468905696e-05,
+      "loss": 0.6577,
+      "step": 2614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808726382321843e-05,
+      "loss": 0.6105,
+      "step": 2615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808508172539832e-05,
+      "loss": 0.6223,
+      "step": 2616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808289839562413e-05,
+      "loss": 0.6415,
+      "step": 2617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980807138339232e-05,
+      "loss": 0.5813,
+      "step": 2618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.5624,
+      "step": 2619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980763410148511e-05,
+      "loss": 0.6525,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807415275753477e-05,
+      "loss": 0.5955,
+      "step": 2621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807196326840163e-05,
+      "loss": 0.6019,
+      "step": 2622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806977254747912e-05,
+      "loss": 0.6365,
+      "step": 2623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806758059479477e-05,
+      "loss": 0.6624,
+      "step": 2624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806538741037614e-05,
+      "loss": 0.6178,
+      "step": 2625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806319299425075e-05,
+      "loss": 0.6043,
+      "step": 2626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806099734644615e-05,
+      "loss": 0.5588,
+      "step": 2627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805880046698995e-05,
+      "loss": 0.5904,
+      "step": 2628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980566023559097e-05,
+      "loss": 0.5921,
+      "step": 2629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805440301323302e-05,
+      "loss": 0.5894,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805220243898756e-05,
+      "loss": 0.5832,
+      "step": 2631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805000063320095e-05,
+      "loss": 0.632,
+      "step": 2632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804779759590083e-05,
+      "loss": 0.6041,
+      "step": 2633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980455933271149e-05,
+      "loss": 0.6833,
+      "step": 2634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804338782687083e-05,
+      "loss": 0.6055,
+      "step": 2635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804118109519634e-05,
+      "loss": 0.5883,
+      "step": 2636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803897313211907e-05,
+      "loss": 0.5966,
+      "step": 2637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803676393766683e-05,
+      "loss": 0.5524,
+      "step": 2638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803455351186732e-05,
+      "loss": 0.5981,
+      "step": 2639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803234185474835e-05,
+      "loss": 0.6139,
+      "step": 2640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980301289663377e-05,
+      "loss": 0.6209,
+      "step": 2641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980279148466631e-05,
+      "loss": 0.6217,
+      "step": 2642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980256994957524e-05,
+      "loss": 0.6292,
+      "step": 2643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980234829136335e-05,
+      "loss": 0.5795,
+      "step": 2644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802126510033407e-05,
+      "loss": 0.5585,
+      "step": 2645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980190460558821e-05,
+      "loss": 0.6387,
+      "step": 2646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980168257803054e-05,
+      "loss": 0.6404,
+      "step": 2647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980146042736319e-05,
+      "loss": 0.6182,
+      "step": 2648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801238153588947e-05,
+      "loss": 0.6085,
+      "step": 2649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801015756710605e-05,
+      "loss": 0.6129,
+      "step": 2650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800793236730956e-05,
+      "loss": 0.667,
+      "step": 2651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800570593652795e-05,
+      "loss": 0.6362,
+      "step": 2652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800347827478917e-05,
+      "loss": 0.5672,
+      "step": 2653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800124938212123e-05,
+      "loss": 0.5983,
+      "step": 2654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979990192585521e-05,
+      "loss": 0.5992,
+      "step": 2655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979967879041098e-05,
+      "loss": 0.6233,
+      "step": 2656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799455531882236e-05,
+      "loss": 0.5743,
+      "step": 2657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799232150271782e-05,
+      "loss": 0.5756,
+      "step": 2658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799008645582424e-05,
+      "loss": 0.5754,
+      "step": 2659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798785017816965e-05,
+      "loss": 0.6089,
+      "step": 2660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979856126697822e-05,
+      "loss": 0.6237,
+      "step": 2661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798337393068998e-05,
+      "loss": 0.5408,
+      "step": 2662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979811339609211e-05,
+      "loss": 0.5957,
+      "step": 2663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797889276050367e-05,
+      "loss": 0.6122,
+      "step": 2664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797665032946586e-05,
+      "loss": 0.5777,
+      "step": 2665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979744066678358e-05,
+      "loss": 0.5631,
+      "step": 2666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797216177564175e-05,
+      "loss": 0.6769,
+      "step": 2667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796991565291187e-05,
+      "loss": 0.5794,
+      "step": 2668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796766829967433e-05,
+      "loss": 0.6449,
+      "step": 2669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979654197159574e-05,
+      "loss": 0.6204,
+      "step": 2670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796316990178933e-05,
+      "loss": 0.5901,
+      "step": 2671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796091885719836e-05,
+      "loss": 0.5701,
+      "step": 2672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795866658221272e-05,
+      "loss": 0.6805,
+      "step": 2673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979564130768608e-05,
+      "loss": 0.6104,
+      "step": 2674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979541583411708e-05,
+      "loss": 0.6022,
+      "step": 2675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979519023751711e-05,
+      "loss": 0.587,
+      "step": 2676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794964517889003e-05,
+      "loss": 0.6198,
+      "step": 2677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794738675235593e-05,
+      "loss": 0.6375,
+      "step": 2678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794512709559715e-05,
+      "loss": 0.6009,
+      "step": 2679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979428662086421e-05,
+      "loss": 0.633,
+      "step": 2680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794060409151914e-05,
+      "loss": 0.6323,
+      "step": 2681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793834074425675e-05,
+      "loss": 0.5872,
+      "step": 2682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979360761668833e-05,
+      "loss": 0.6,
+      "step": 2683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793381035942722e-05,
+      "loss": 0.6838,
+      "step": 2684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793154332191702e-05,
+      "loss": 0.6452,
+      "step": 2685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792927505438114e-05,
+      "loss": 0.5628,
+      "step": 2686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792700555684812e-05,
+      "loss": 0.6401,
+      "step": 2687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979247348293464e-05,
+      "loss": 0.5864,
+      "step": 2688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979224628719045e-05,
+      "loss": 0.716,
+      "step": 2689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792018968455102e-05,
+      "loss": 0.5915,
+      "step": 2690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791791526731444e-05,
+      "loss": 0.6687,
+      "step": 2691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979156396202234e-05,
+      "loss": 0.6531,
+      "step": 2692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979133627433064e-05,
+      "loss": 0.6233,
+      "step": 2693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979110846365921e-05,
+      "loss": 0.5373,
+      "step": 2694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790880530010913e-05,
+      "loss": 0.6215,
+      "step": 2695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97906524733886e-05,
+      "loss": 0.6546,
+      "step": 2696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790424293795148e-05,
+      "loss": 0.6381,
+      "step": 2697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979019599123342e-05,
+      "loss": 0.6439,
+      "step": 2698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978996756570628e-05,
+      "loss": 0.5684,
+      "step": 2699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97897390172166e-05,
+      "loss": 0.5938,
+      "step": 2700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978951034576725e-05,
+      "loss": 0.5696,
+      "step": 2701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789281551361103e-05,
+      "loss": 0.5991,
+      "step": 2702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978905263400103e-05,
+      "loss": 0.6213,
+      "step": 2703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788823593689908e-05,
+      "loss": 0.6366,
+      "step": 2704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788594430430613e-05,
+      "loss": 0.6271,
+      "step": 2705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788365144226024e-05,
+      "loss": 0.5939,
+      "step": 2706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978813573507902e-05,
+      "loss": 0.5451,
+      "step": 2707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787906202992483e-05,
+      "loss": 0.5959,
+      "step": 2708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787676547969295e-05,
+      "loss": 0.6464,
+      "step": 2709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787446770012342e-05,
+      "loss": 0.6296,
+      "step": 2710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978721686912451e-05,
+      "loss": 0.624,
+      "step": 2711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786986845308687e-05,
+      "loss": 0.6805,
+      "step": 2712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786756698567758e-05,
+      "loss": 0.5789,
+      "step": 2713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978652642890462e-05,
+      "loss": 0.5384,
+      "step": 2714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978629603632216e-05,
+      "loss": 0.5744,
+      "step": 2715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786065520823274e-05,
+      "loss": 0.6195,
+      "step": 2716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978583488241086e-05,
+      "loss": 0.5685,
+      "step": 2717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978560412108781e-05,
+      "loss": 0.5816,
+      "step": 2718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978537323685702e-05,
+      "loss": 0.6062,
+      "step": 2719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785142229721396e-05,
+      "loss": 0.5547,
+      "step": 2720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978491109968384e-05,
+      "loss": 0.6268,
+      "step": 2721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784679846747253e-05,
+      "loss": 0.6084,
+      "step": 2722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784448470914538e-05,
+      "loss": 0.6673,
+      "step": 2723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97842169721886e-05,
+      "loss": 0.5839,
+      "step": 2724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978398535057235e-05,
+      "loss": 0.5718,
+      "step": 2725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97837536060687e-05,
+      "loss": 0.6107,
+      "step": 2726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783521738680552e-05,
+      "loss": 0.6626,
+      "step": 2727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783289748410825e-05,
+      "loss": 0.5651,
+      "step": 2728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978305763526243e-05,
+      "loss": 0.6812,
+      "step": 2729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782825399238285e-05,
+      "loss": 0.594,
+      "step": 2730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97825930403413e-05,
+      "loss": 0.6032,
+      "step": 2731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782360558574406e-05,
+      "loss": 0.5937,
+      "step": 2732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782127953940508e-05,
+      "loss": 0.633,
+      "step": 2733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978189522644254e-05,
+      "loss": 0.5902,
+      "step": 2734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781662376083418e-05,
+      "loss": 0.5587,
+      "step": 2735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978142940286607e-05,
+      "loss": 0.5984,
+      "step": 2736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978119630679342e-05,
+      "loss": 0.5686,
+      "step": 2737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780963087868397e-05,
+      "loss": 0.6145,
+      "step": 2738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978072974609393e-05,
+      "loss": 0.585,
+      "step": 2739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780496281472952e-05,
+      "loss": 0.5779,
+      "step": 2740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978026269400839e-05,
+      "loss": 0.578,
+      "step": 2741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780028983703177e-05,
+      "loss": 0.6547,
+      "step": 2742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977979515056026e-05,
+      "loss": 0.6661,
+      "step": 2743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779561194582563e-05,
+      "loss": 0.5723,
+      "step": 2744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779327115773032e-05,
+      "loss": 0.6203,
+      "step": 2745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779092914134606e-05,
+      "loss": 0.6063,
+      "step": 2746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778858589670223e-05,
+      "loss": 0.5722,
+      "step": 2747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778624142382826e-05,
+      "loss": 0.6706,
+      "step": 2748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778389572275368e-05,
+      "loss": 0.6262,
+      "step": 2749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778154879350785e-05,
+      "loss": 0.5744,
+      "step": 2750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977792006361203e-05,
+      "loss": 0.6445,
+      "step": 2751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977768512506205e-05,
+      "loss": 0.5859,
+      "step": 2752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97774500637038e-05,
+      "loss": 0.6051,
+      "step": 2753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977721487954023e-05,
+      "loss": 0.6701,
+      "step": 2754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776979572574292e-05,
+      "loss": 0.6619,
+      "step": 2755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776744142808942e-05,
+      "loss": 0.5576,
+      "step": 2756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776508590247138e-05,
+      "loss": 0.5475,
+      "step": 2757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776272914891842e-05,
+      "loss": 0.5761,
+      "step": 2758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776037116746005e-05,
+      "loss": 0.6349,
+      "step": 2759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97758011958126e-05,
+      "loss": 0.5885,
+      "step": 2760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977556515209458e-05,
+      "loss": 0.6282,
+      "step": 2761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775328985594913e-05,
+      "loss": 0.5528,
+      "step": 2762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977509269631657e-05,
+      "loss": 0.6029,
+      "step": 2763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774856284262518e-05,
+      "loss": 0.5772,
+      "step": 2764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977461974943572e-05,
+      "loss": 0.6734,
+      "step": 2765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774383091839153e-05,
+      "loss": 0.5634,
+      "step": 2766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774146311475784e-05,
+      "loss": 0.6331,
+      "step": 2767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773909408348592e-05,
+      "loss": 0.5306,
+      "step": 2768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773672382460546e-05,
+      "loss": 0.5613,
+      "step": 2769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977343523381463e-05,
+      "loss": 0.5868,
+      "step": 2770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773197962413825e-05,
+      "loss": 0.5964,
+      "step": 2771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97729605682611e-05,
+      "loss": 0.6143,
+      "step": 2772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772723051359444e-05,
+      "loss": 0.6809,
+      "step": 2773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772485411711843e-05,
+      "loss": 0.6512,
+      "step": 2774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772247649321273e-05,
+      "loss": 0.7041,
+      "step": 2775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772009764190727e-05,
+      "loss": 0.5787,
+      "step": 2776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771771756323193e-05,
+      "loss": 0.5949,
+      "step": 2777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977153362572166e-05,
+      "loss": 0.6077,
+      "step": 2778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977129537238911e-05,
+      "loss": 0.5789,
+      "step": 2779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977105699632855e-05,
+      "loss": 0.5926,
+      "step": 2780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770818497542962e-05,
+      "loss": 0.5312,
+      "step": 2781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770579876035348e-05,
+      "loss": 0.6025,
+      "step": 2782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770341131808703e-05,
+      "loss": 0.5731,
+      "step": 2783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770102264866028e-05,
+      "loss": 0.6656,
+      "step": 2784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976986327521032e-05,
+      "loss": 0.6658,
+      "step": 2785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976962416284458e-05,
+      "loss": 0.5624,
+      "step": 2786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976938492777182e-05,
+      "loss": 0.5839,
+      "step": 2787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769145569995033e-05,
+      "loss": 0.6895,
+      "step": 2788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976890608951723e-05,
+      "loss": 0.586,
+      "step": 2789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768666486341422e-05,
+      "loss": 0.6778,
+      "step": 2790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768426760470615e-05,
+      "loss": 0.6158,
+      "step": 2791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976818691190782e-05,
+      "loss": 0.6518,
+      "step": 2792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976794694065605e-05,
+      "loss": 0.5793,
+      "step": 2793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767706846718322e-05,
+      "loss": 0.5654,
+      "step": 2794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767466630097645e-05,
+      "loss": 0.6324,
+      "step": 2795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767226290797047e-05,
+      "loss": 0.5859,
+      "step": 2796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766985828819535e-05,
+      "loss": 0.6143,
+      "step": 2797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766745244168135e-05,
+      "loss": 0.6373,
+      "step": 2798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766504536845862e-05,
+      "loss": 0.6097,
+      "step": 2799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766263706855753e-05,
+      "loss": 0.6919,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976602275420082e-05,
+      "loss": 0.6132,
+      "step": 2801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765781678884094e-05,
+      "loss": 0.6361,
+      "step": 2802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765540480908606e-05,
+      "loss": 0.5881,
+      "step": 2803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765299160277383e-05,
+      "loss": 0.5705,
+      "step": 2804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765057716993455e-05,
+      "loss": 0.5724,
+      "step": 2805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764816151059854e-05,
+      "loss": 0.6234,
+      "step": 2806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764574462479616e-05,
+      "loss": 0.6039,
+      "step": 2807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764332651255777e-05,
+      "loss": 0.6093,
+      "step": 2808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976409071739137e-05,
+      "loss": 0.6598,
+      "step": 2809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763848660889437e-05,
+      "loss": 0.6749,
+      "step": 2810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763606481753023e-05,
+      "loss": 0.5984,
+      "step": 2811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976336417998516e-05,
+      "loss": 0.6079,
+      "step": 2812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763121755588898e-05,
+      "loss": 0.5852,
+      "step": 2813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762879208567277e-05,
+      "loss": 0.6603,
+      "step": 2814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976263653892335e-05,
+      "loss": 0.6491,
+      "step": 2815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976239374666016e-05,
+      "loss": 0.6058,
+      "step": 2816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762150831780755e-05,
+      "loss": 0.6168,
+      "step": 2817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761907794288193e-05,
+      "loss": 0.5883,
+      "step": 2818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761664634185523e-05,
+      "loss": 0.5796,
+      "step": 2819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761421351475794e-05,
+      "loss": 0.6459,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976117794616207e-05,
+      "loss": 0.592,
+      "step": 2821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760934418247402e-05,
+      "loss": 0.6299,
+      "step": 2822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760690767734852e-05,
+      "loss": 0.5644,
+      "step": 2823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760446994627477e-05,
+      "loss": 0.5721,
+      "step": 2824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760203098928345e-05,
+      "loss": 0.6188,
+      "step": 2825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759959080640513e-05,
+      "loss": 0.6328,
+      "step": 2826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975971493976705e-05,
+      "loss": 0.5773,
+      "step": 2827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759470676311018e-05,
+      "loss": 0.538,
+      "step": 2828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975922629027549e-05,
+      "loss": 0.6082,
+      "step": 2829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975898178166353e-05,
+      "loss": 0.5647,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975873715047822e-05,
+      "loss": 0.5856,
+      "step": 2831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758492396722618e-05,
+      "loss": 0.6004,
+      "step": 2832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758247520399806e-05,
+      "loss": 0.6046,
+      "step": 2833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975800252151286e-05,
+      "loss": 0.6644,
+      "step": 2834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757757400064857e-05,
+      "loss": 0.5844,
+      "step": 2835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757512156058873e-05,
+      "loss": 0.597,
+      "step": 2836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757266789497985e-05,
+      "loss": 0.5789,
+      "step": 2837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757021300385288e-05,
+      "loss": 0.6263,
+      "step": 2838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756775688723853e-05,
+      "loss": 0.5938,
+      "step": 2839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975652995451677e-05,
+      "loss": 0.6118,
+      "step": 2840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756284097767123e-05,
+      "loss": 0.6011,
+      "step": 2841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756038118478e-05,
+      "loss": 0.5874,
+      "step": 2842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755792016652495e-05,
+      "loss": 0.5618,
+      "step": 2843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755545792293694e-05,
+      "loss": 0.5227,
+      "step": 2844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755299445404694e-05,
+      "loss": 0.6587,
+      "step": 2845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755052975988582e-05,
+      "loss": 0.5909,
+      "step": 2846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975480638404846e-05,
+      "loss": 0.5575,
+      "step": 2847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975455966958743e-05,
+      "loss": 0.6052,
+      "step": 2848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754312832608577e-05,
+      "loss": 0.5453,
+      "step": 2849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754065873115008e-05,
+      "loss": 0.5066,
+      "step": 2850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975381879110983e-05,
+      "loss": 0.6469,
+      "step": 2851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753571586596138e-05,
+      "loss": 0.6104,
+      "step": 2852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975332425957704e-05,
+      "loss": 0.528,
+      "step": 2853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753076810055646e-05,
+      "loss": 0.5867,
+      "step": 2854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975282923803506e-05,
+      "loss": 0.6148,
+      "step": 2855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752581543518392e-05,
+      "loss": 0.6265,
+      "step": 2856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752333726508755e-05,
+      "loss": 0.5813,
+      "step": 2857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752085787009258e-05,
+      "loss": 0.6472,
+      "step": 2858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975183772502302e-05,
+      "loss": 0.6081,
+      "step": 2859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975158954055315e-05,
+      "loss": 0.6285,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751341233602773e-05,
+      "loss": 0.5778,
+      "step": 2861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751092804175004e-05,
+      "loss": 0.5976,
+      "step": 2862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975084425227296e-05,
+      "loss": 0.6368,
+      "step": 2863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750595577899768e-05,
+      "loss": 0.5849,
+      "step": 2864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750346781058553e-05,
+      "loss": 0.5732,
+      "step": 2865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975009786175243e-05,
+      "loss": 0.6479,
+      "step": 2866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749848819984536e-05,
+      "loss": 0.6608,
+      "step": 2867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749599655757992e-05,
+      "loss": 0.5537,
+      "step": 2868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749350369075932e-05,
+      "loss": 0.6018,
+      "step": 2869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749100959941484e-05,
+      "loss": 0.6618,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748851428357785e-05,
+      "loss": 0.5782,
+      "step": 2871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974860177432796e-05,
+      "loss": 0.5977,
+      "step": 2872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974835199785516e-05,
+      "loss": 0.5329,
+      "step": 2873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974810209894251e-05,
+      "loss": 0.5916,
+      "step": 2874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747852077593147e-05,
+      "loss": 0.5845,
+      "step": 2875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974760193381022e-05,
+      "loss": 0.6306,
+      "step": 2876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974735166759686e-05,
+      "loss": 0.5973,
+      "step": 2877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747101278956226e-05,
+      "loss": 0.6294,
+      "step": 2878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974685076789145e-05,
+      "loss": 0.5788,
+      "step": 2879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746600134405682e-05,
+      "loss": 0.5901,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974634937850207e-05,
+      "loss": 0.5276,
+      "step": 2881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974609850018377e-05,
+      "loss": 0.6409,
+      "step": 2882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745847499453918e-05,
+      "loss": 0.6195,
+      "step": 2883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974559637631568e-05,
+      "loss": 0.6177,
+      "step": 2884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745345130772206e-05,
+      "loss": 0.55,
+      "step": 2885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974509376282665e-05,
+      "loss": 0.5826,
+      "step": 2886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974484227248217e-05,
+      "loss": 0.6592,
+      "step": 2887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744590659741928e-05,
+      "loss": 0.5922,
+      "step": 2888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974433892460908e-05,
+      "loss": 0.6649,
+      "step": 2889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744087067086792e-05,
+      "loss": 0.5685,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974383508717822e-05,
+      "loss": 0.5617,
+      "step": 2891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974358298488654e-05,
+      "loss": 0.5572,
+      "step": 2892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743330760214906e-05,
+      "loss": 0.5852,
+      "step": 2893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743078413166494e-05,
+      "loss": 0.6297,
+      "step": 2894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974282594374447e-05,
+      "loss": 0.5814,
+      "step": 2895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742573351952008e-05,
+      "loss": 0.5836,
+      "step": 2896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974232063779228e-05,
+      "loss": 0.6342,
+      "step": 2897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742067801268457e-05,
+      "loss": 0.6514,
+      "step": 2898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974181484238372e-05,
+      "loss": 0.605,
+      "step": 2899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974156176114124e-05,
+      "loss": 0.5843,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97413085575442e-05,
+      "loss": 0.5992,
+      "step": 2901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974105523159578e-05,
+      "loss": 0.6555,
+      "step": 2902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740801783299158e-05,
+      "loss": 0.5721,
+      "step": 2903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974054821265752e-05,
+      "loss": 0.573,
+      "step": 2904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974029451967405e-05,
+      "loss": 0.544,
+      "step": 2905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740040704351938e-05,
+      "loss": 0.5897,
+      "step": 2906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739786766694368e-05,
+      "loss": 0.5738,
+      "step": 2907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739532706704535e-05,
+      "loss": 0.6559,
+      "step": 2908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973927852438562e-05,
+      "loss": 0.665,
+      "step": 2909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739024219740822e-05,
+      "loss": 0.6169,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738769792773338e-05,
+      "loss": 0.5804,
+      "step": 2911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738515243486358e-05,
+      "loss": 0.643,
+      "step": 2912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973826057188308e-05,
+      "loss": 0.5713,
+      "step": 2913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738005777966704e-05,
+      "loss": 0.5968,
+      "step": 2914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737750861740434e-05,
+      "loss": 0.625,
+      "step": 2915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973749582320746e-05,
+      "loss": 0.6186,
+      "step": 2916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737240662371004e-05,
+      "loss": 0.5832,
+      "step": 2917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736985379234252e-05,
+      "loss": 0.5857,
+      "step": 2918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973672997380042e-05,
+      "loss": 0.5923,
+      "step": 2919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736474446072712e-05,
+      "loss": 0.5909,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973621879605434e-05,
+      "loss": 0.6441,
+      "step": 2921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735963023748518e-05,
+      "loss": 0.5696,
+      "step": 2922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973570712915845e-05,
+      "loss": 0.5553,
+      "step": 2923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735451112287363e-05,
+      "loss": 0.6035,
+      "step": 2924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973519497313846e-05,
+      "loss": 0.5453,
+      "step": 2925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734938711714965e-05,
+      "loss": 0.5637,
+      "step": 2926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734682328020092e-05,
+      "loss": 0.5972,
+      "step": 2927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734425822057063e-05,
+      "loss": 0.5862,
+      "step": 2928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734169193829104e-05,
+      "loss": 0.6199,
+      "step": 2929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733912443339434e-05,
+      "loss": 0.5141,
+      "step": 2930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733655570591277e-05,
+      "loss": 0.5839,
+      "step": 2931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973339857558786e-05,
+      "loss": 0.613,
+      "step": 2932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733141458332416e-05,
+      "loss": 0.5381,
+      "step": 2933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732884218828165e-05,
+      "loss": 0.5844,
+      "step": 2934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732626857078347e-05,
+      "loss": 0.6405,
+      "step": 2935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732369373086188e-05,
+      "loss": 0.5967,
+      "step": 2936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732111766854926e-05,
+      "loss": 0.5581,
+      "step": 2937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9731854038387795e-05,
+      "loss": 0.5492,
+      "step": 2938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731596187688032e-05,
+      "loss": 0.5981,
+      "step": 2939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731338214758874e-05,
+      "loss": 0.5987,
+      "step": 2940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731080119603566e-05,
+      "loss": 0.5901,
+      "step": 2941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730821902225345e-05,
+      "loss": 0.6483,
+      "step": 2942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973056356262746e-05,
+      "loss": 0.5219,
+      "step": 2943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730305100813147e-05,
+      "loss": 0.5737,
+      "step": 2944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973004651678566e-05,
+      "loss": 0.6233,
+      "step": 2945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972978781054824e-05,
+      "loss": 0.5813,
+      "step": 2946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972952898210414e-05,
+      "loss": 0.6441,
+      "step": 2947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729270031456614e-05,
+      "loss": 0.6765,
+      "step": 2948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729010958608912e-05,
+      "loss": 0.6242,
+      "step": 2949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728751763564284e-05,
+      "loss": 0.6175,
+      "step": 2950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728492446325994e-05,
+      "loss": 0.6136,
+      "step": 2951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972823300689729e-05,
+      "loss": 0.535,
+      "step": 2952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727973445281433e-05,
+      "loss": 0.567,
+      "step": 2953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972771376148169e-05,
+      "loss": 0.6382,
+      "step": 2954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727453955501315e-05,
+      "loss": 0.5661,
+      "step": 2955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727194027343572e-05,
+      "loss": 0.5818,
+      "step": 2956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972693397701173e-05,
+      "loss": 0.5418,
+      "step": 2957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726673804509052e-05,
+      "loss": 0.5613,
+      "step": 2958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726413509838804e-05,
+      "loss": 0.6552,
+      "step": 2959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726153093004257e-05,
+      "loss": 0.5968,
+      "step": 2960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725892554008687e-05,
+      "loss": 0.6279,
+      "step": 2961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972563189285536e-05,
+      "loss": 0.6012,
+      "step": 2962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972537110954755e-05,
+      "loss": 0.644,
+      "step": 2963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725110204088532e-05,
+      "loss": 0.6059,
+      "step": 2964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972484917648159e-05,
+      "loss": 0.6304,
+      "step": 2965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972458802673e-05,
+      "loss": 0.6714,
+      "step": 2966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724326754837034e-05,
+      "loss": 0.5869,
+      "step": 2967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972406536080598e-05,
+      "loss": 0.5839,
+      "step": 2968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972380384464012e-05,
+      "loss": 0.6358,
+      "step": 2969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723542206342742e-05,
+      "loss": 0.5755,
+      "step": 2970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723280445917127e-05,
+      "loss": 0.6915,
+      "step": 2971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723018563366567e-05,
+      "loss": 0.551,
+      "step": 2972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722756558694346e-05,
+      "loss": 0.5529,
+      "step": 2973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722494431903763e-05,
+      "loss": 0.5843,
+      "step": 2974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97222321829981e-05,
+      "loss": 0.5596,
+      "step": 2975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721969811980658e-05,
+      "loss": 0.6627,
+      "step": 2976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721707318854735e-05,
+      "loss": 0.5975,
+      "step": 2977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972144470362362e-05,
+      "loss": 0.6042,
+      "step": 2978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721181966290614e-05,
+      "loss": 0.6373,
+      "step": 2979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972091910685902e-05,
+      "loss": 0.6043,
+      "step": 2980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720656125332137e-05,
+      "loss": 0.5815,
+      "step": 2981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972039302171327e-05,
+      "loss": 0.6662,
+      "step": 2982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972012979600572e-05,
+      "loss": 0.6,
+      "step": 2983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719866448212794e-05,
+      "loss": 0.6466,
+      "step": 2984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719602978337805e-05,
+      "loss": 0.6053,
+      "step": 2985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719339386384056e-05,
+      "loss": 0.589,
+      "step": 2986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719075672354862e-05,
+      "loss": 0.5938,
+      "step": 2987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971881183625353e-05,
+      "loss": 0.5838,
+      "step": 2988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971854787808338e-05,
+      "loss": 0.5269,
+      "step": 2989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971828379784772e-05,
+      "loss": 0.5591,
+      "step": 2990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718019595549877e-05,
+      "loss": 0.5211,
+      "step": 2991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971775527119316e-05,
+      "loss": 0.6405,
+      "step": 2992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717490824780895e-05,
+      "loss": 0.5748,
+      "step": 2993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717226256316398e-05,
+      "loss": 0.5293,
+      "step": 2994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716961565802997e-05,
+      "loss": 0.5761,
+      "step": 2995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716696753244014e-05,
+      "loss": 0.5441,
+      "step": 2996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716431818642776e-05,
+      "loss": 0.5992,
+      "step": 2997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971616676200261e-05,
+      "loss": 0.5576,
+      "step": 2998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715901583326845e-05,
+      "loss": 0.6117,
+      "step": 2999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971563628261881e-05,
+      "loss": 0.5807,
+      "step": 3000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715370859881843e-05,
+      "loss": 0.594,
+      "step": 3001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715105315119273e-05,
+      "loss": 0.5392,
+      "step": 3002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714839648334434e-05,
+      "loss": 0.6177,
+      "step": 3003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714573859530666e-05,
+      "loss": 0.5724,
+      "step": 3004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714307948711305e-05,
+      "loss": 0.607,
+      "step": 3005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714041915879694e-05,
+      "loss": 0.6088,
+      "step": 3006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971377576103917e-05,
+      "loss": 0.5857,
+      "step": 3007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713509484193084e-05,
+      "loss": 0.6258,
+      "step": 3008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971324308534477e-05,
+      "loss": 0.6226,
+      "step": 3009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712976564497583e-05,
+      "loss": 0.631,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971270992165486e-05,
+      "loss": 0.6133,
+      "step": 3011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971244315681996e-05,
+      "loss": 0.5543,
+      "step": 3012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971217626999623e-05,
+      "loss": 0.6179,
+      "step": 3013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711909261187022e-05,
+      "loss": 0.6516,
+      "step": 3014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971164213039569e-05,
+      "loss": 0.6992,
+      "step": 3015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971137487762559e-05,
+      "loss": 0.5541,
+      "step": 3016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711107502880074e-05,
+      "loss": 0.5607,
+      "step": 3017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710840006162507e-05,
+      "loss": 0.581,
+      "step": 3018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710572387476246e-05,
+      "loss": 0.553,
+      "step": 3019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971030464682465e-05,
+      "loss": 0.5834,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710036784211083e-05,
+      "loss": 0.6471,
+      "step": 3021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970976879963891e-05,
+      "loss": 0.5451,
+      "step": 3022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97095006931115e-05,
+      "loss": 0.6166,
+      "step": 3023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709232464632217e-05,
+      "loss": 0.6732,
+      "step": 3024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970896411420443e-05,
+      "loss": 0.6104,
+      "step": 3025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970869564183151e-05,
+      "loss": 0.5129,
+      "step": 3026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708427047516827e-05,
+      "loss": 0.6353,
+      "step": 3027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970815833126376e-05,
+      "loss": 0.659,
+      "step": 3028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707889493075678e-05,
+      "loss": 0.627,
+      "step": 3029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707620532955962e-05,
+      "loss": 0.6001,
+      "step": 3030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707351450907987e-05,
+      "loss": 0.6621,
+      "step": 3031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707082246935133e-05,
+      "loss": 0.6226,
+      "step": 3032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706812921040786e-05,
+      "loss": 0.5737,
+      "step": 3033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706543473228322e-05,
+      "loss": 0.5793,
+      "step": 3034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970627390350113e-05,
+      "loss": 0.5682,
+      "step": 3035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706004211862596e-05,
+      "loss": 0.5926,
+      "step": 3036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97057343983161e-05,
+      "loss": 0.5428,
+      "step": 3037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970546446286504e-05,
+      "loss": 0.5415,
+      "step": 3038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705194405512805e-05,
+      "loss": 0.6157,
+      "step": 3039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970492422626278e-05,
+      "loss": 0.5863,
+      "step": 3040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704653925118363e-05,
+      "loss": 0.5759,
+      "step": 3041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704383502082954e-05,
+      "loss": 0.5245,
+      "step": 3042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970411295715994e-05,
+      "loss": 0.6018,
+      "step": 3043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703842290352727e-05,
+      "loss": 0.586,
+      "step": 3044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970357150166471e-05,
+      "loss": 0.5545,
+      "step": 3045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970330059109929e-05,
+      "loss": 0.6675,
+      "step": 3046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703029558659872e-05,
+      "loss": 0.5951,
+      "step": 3047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702758404349862e-05,
+      "loss": 0.621,
+      "step": 3048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970248712817266e-05,
+      "loss": 0.6255,
+      "step": 3049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702215730131677e-05,
+      "loss": 0.6996,
+      "step": 3050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701944210230324e-05,
+      "loss": 0.6058,
+      "step": 3051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701672568472005e-05,
+      "loss": 0.5202,
+      "step": 3052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701400804860136e-05,
+      "loss": 0.6069,
+      "step": 3053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970112891939813e-05,
+      "loss": 0.648,
+      "step": 3054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97008569120894e-05,
+      "loss": 0.5531,
+      "step": 3055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700584782937366e-05,
+      "loss": 0.6037,
+      "step": 3056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700312531945444e-05,
+      "loss": 0.5911,
+      "step": 3057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700040159117047e-05,
+      "loss": 0.6217,
+      "step": 3058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969976766445561e-05,
+      "loss": 0.7725,
+      "step": 3059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699495047964545e-05,
+      "loss": 0.632,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969922230964728e-05,
+      "loss": 0.6015,
+      "step": 3061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698949449507235e-05,
+      "loss": 0.6125,
+      "step": 3062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698676467547848e-05,
+      "loss": 0.5328,
+      "step": 3063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969840336377254e-05,
+      "loss": 0.6212,
+      "step": 3064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969813013818474e-05,
+      "loss": 0.6556,
+      "step": 3065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697856790787882e-05,
+      "loss": 0.5809,
+      "step": 3066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697583321585406e-05,
+      "loss": 0.5264,
+      "step": 3067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697309730580733e-05,
+      "loss": 0.6334,
+      "step": 3068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969703601777731e-05,
+      "loss": 0.6149,
+      "step": 3069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969676218317857e-05,
+      "loss": 0.5853,
+      "step": 3070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696488226787956e-05,
+      "loss": 0.5182,
+      "step": 3071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696214148608903e-05,
+      "loss": 0.5528,
+      "step": 3072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969593994864486e-05,
+      "loss": 0.6327,
+      "step": 3073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969566562689927e-05,
+      "loss": 0.5851,
+      "step": 3074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695391183375575e-05,
+      "loss": 0.6161,
+      "step": 3075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695116618077222e-05,
+      "loss": 0.5881,
+      "step": 3076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694841931007664e-05,
+      "loss": 0.6041,
+      "step": 3077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694567122170346e-05,
+      "loss": 0.5723,
+      "step": 3078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694292191568722e-05,
+      "loss": 0.5656,
+      "step": 3079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694017139206248e-05,
+      "loss": 0.581,
+      "step": 3080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969374196508637e-05,
+      "loss": 0.5911,
+      "step": 3081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693466669212557e-05,
+      "loss": 0.5715,
+      "step": 3082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969319125158826e-05,
+      "loss": 0.6315,
+      "step": 3083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692915712216934e-05,
+      "loss": 0.5927,
+      "step": 3084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692640051102044e-05,
+      "loss": 0.7064,
+      "step": 3085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969236426824705e-05,
+      "loss": 0.5748,
+      "step": 3086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692088363655423e-05,
+      "loss": 0.5971,
+      "step": 3087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969181233733062e-05,
+      "loss": 0.6747,
+      "step": 3088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691536189276113e-05,
+      "loss": 0.5853,
+      "step": 3089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691259919495366e-05,
+      "loss": 0.6664,
+      "step": 3090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690983527991854e-05,
+      "loss": 0.6585,
+      "step": 3091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690707014769046e-05,
+      "loss": 0.6156,
+      "step": 3092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690430379830413e-05,
+      "loss": 0.6318,
+      "step": 3093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690153623179437e-05,
+      "loss": 0.6029,
+      "step": 3094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968987674481958e-05,
+      "loss": 0.62,
+      "step": 3095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689599744754338e-05,
+      "loss": 0.5498,
+      "step": 3096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689322622987178e-05,
+      "loss": 0.6657,
+      "step": 3097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968904537952158e-05,
+      "loss": 0.5705,
+      "step": 3098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968876801436103e-05,
+      "loss": 0.5219,
+      "step": 3099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968849052750901e-05,
+      "loss": 0.64,
+      "step": 3100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968821291896901e-05,
+      "loss": 0.6475,
+      "step": 3101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968793518874451e-05,
+      "loss": 0.5934,
+      "step": 3102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687657336839003e-05,
+      "loss": 0.5952,
+      "step": 3103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687379363255978e-05,
+      "loss": 0.5367,
+      "step": 3104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687101267998923e-05,
+      "loss": 0.5641,
+      "step": 3105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968682305107134e-05,
+      "loss": 0.6132,
+      "step": 3106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968654471247671e-05,
+      "loss": 0.5629,
+      "step": 3107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686266252218536e-05,
+      "loss": 0.5926,
+      "step": 3108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968598767030032e-05,
+      "loss": 0.6053,
+      "step": 3109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685708966725552e-05,
+      "loss": 0.5927,
+      "step": 3110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968543014149774e-05,
+      "loss": 0.6215,
+      "step": 3111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685151194620384e-05,
+      "loss": 0.6833,
+      "step": 3112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968487212609698e-05,
+      "loss": 0.5116,
+      "step": 3113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684592935931047e-05,
+      "loss": 0.5512,
+      "step": 3114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684313624126083e-05,
+      "loss": 0.6173,
+      "step": 3115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684034190685595e-05,
+      "loss": 0.6302,
+      "step": 3116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683754635613094e-05,
+      "loss": 0.5844,
+      "step": 3117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683474958912096e-05,
+      "loss": 0.5857,
+      "step": 3118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683195160586108e-05,
+      "loss": 0.5385,
+      "step": 3119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682915240638647e-05,
+      "loss": 0.5723,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968263519907323e-05,
+      "loss": 0.599,
+      "step": 3121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968235503589337e-05,
+      "loss": 0.6466,
+      "step": 3122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682074751102588e-05,
+      "loss": 0.6568,
+      "step": 3123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681794344704407e-05,
+      "loss": 0.5669,
+      "step": 3124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681513816702348e-05,
+      "loss": 0.5807,
+      "step": 3125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968123316709993e-05,
+      "loss": 0.5832,
+      "step": 3126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680952395900684e-05,
+      "loss": 0.5724,
+      "step": 3127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968067150310813e-05,
+      "loss": 0.6286,
+      "step": 3128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680390488725805e-05,
+      "loss": 0.5974,
+      "step": 3129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680109352757228e-05,
+      "loss": 0.6113,
+      "step": 3130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679828095205938e-05,
+      "loss": 0.5484,
+      "step": 3131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679546716075463e-05,
+      "loss": 0.6253,
+      "step": 3132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679265215369342e-05,
+      "loss": 0.5728,
+      "step": 3133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678983593091107e-05,
+      "loss": 0.5611,
+      "step": 3134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678701849244297e-05,
+      "loss": 0.6233,
+      "step": 3135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678419983832448e-05,
+      "loss": 0.5595,
+      "step": 3136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96781379968591e-05,
+      "loss": 0.5968,
+      "step": 3137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677855888327803e-05,
+      "loss": 0.5776,
+      "step": 3138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967757365824209e-05,
+      "loss": 0.64,
+      "step": 3139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677291306605507e-05,
+      "loss": 0.5766,
+      "step": 3140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677008833421606e-05,
+      "loss": 0.6781,
+      "step": 3141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676726238693934e-05,
+      "loss": 0.5983,
+      "step": 3142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676443522426037e-05,
+      "loss": 0.5768,
+      "step": 3143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676160684621468e-05,
+      "loss": 0.6092,
+      "step": 3144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675877725283775e-05,
+      "loss": 0.5811,
+      "step": 3145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967559464441652e-05,
+      "loss": 0.5209,
+      "step": 3146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675311442023254e-05,
+      "loss": 0.5629,
+      "step": 3147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675028118107534e-05,
+      "loss": 0.6307,
+      "step": 3148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967474467267292e-05,
+      "loss": 0.6298,
+      "step": 3149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674461105722966e-05,
+      "loss": 0.6673,
+      "step": 3150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674177417261243e-05,
+      "loss": 0.6147,
+      "step": 3151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967389360729131e-05,
+      "loss": 0.6194,
+      "step": 3152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673609675816728e-05,
+      "loss": 0.6234,
+      "step": 3153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967332562284107e-05,
+      "loss": 0.5706,
+      "step": 3154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673041448367904e-05,
+      "loss": 0.6063,
+      "step": 3155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967275715240079e-05,
+      "loss": 0.6946,
+      "step": 3156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672472734943304e-05,
+      "loss": 0.6186,
+      "step": 3157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967218819599902e-05,
+      "loss": 0.5636,
+      "step": 3158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967190353557151e-05,
+      "loss": 0.6002,
+      "step": 3159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671618753664355e-05,
+      "loss": 0.5826,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967133385028112e-05,
+      "loss": 0.6296,
+      "step": 3161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671048825425397e-05,
+      "loss": 0.6193,
+      "step": 3162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670763679100754e-05,
+      "loss": 0.6058,
+      "step": 3163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670478411310777e-05,
+      "loss": 0.6204,
+      "step": 3164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670193022059053e-05,
+      "loss": 0.6031,
+      "step": 3165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966990751134916e-05,
+      "loss": 0.6249,
+      "step": 3166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669621879184688e-05,
+      "loss": 0.6081,
+      "step": 3167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669336125569226e-05,
+      "loss": 0.6325,
+      "step": 3168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669050250506356e-05,
+      "loss": 0.5167,
+      "step": 3169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668764253999677e-05,
+      "loss": 0.5978,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668478136052776e-05,
+      "loss": 0.6081,
+      "step": 3171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668191896669252e-05,
+      "loss": 0.5657,
+      "step": 3172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966790553585269e-05,
+      "loss": 0.539,
+      "step": 3173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667619053606696e-05,
+      "loss": 0.5562,
+      "step": 3174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667332449934867e-05,
+      "loss": 0.6622,
+      "step": 3175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667045724840797e-05,
+      "loss": 0.6214,
+      "step": 3176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666758878328095e-05,
+      "loss": 0.6163,
+      "step": 3177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966647191040036e-05,
+      "loss": 0.6182,
+      "step": 3178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966618482106119e-05,
+      "loss": 0.5957,
+      "step": 3179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665897610314207e-05,
+      "loss": 0.5992,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665610278163002e-05,
+      "loss": 0.6146,
+      "step": 3181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665322824611192e-05,
+      "loss": 0.5627,
+      "step": 3182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665035249662387e-05,
+      "loss": 0.549,
+      "step": 3183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664747553320197e-05,
+      "loss": 0.6026,
+      "step": 3184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966445973558824e-05,
+      "loss": 0.5832,
+      "step": 3185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664171796470123e-05,
+      "loss": 0.5912,
+      "step": 3186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663883735969465e-05,
+      "loss": 0.6548,
+      "step": 3187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966359555408989e-05,
+      "loss": 0.5588,
+      "step": 3188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663307250835015e-05,
+      "loss": 0.6649,
+      "step": 3189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663018826208456e-05,
+      "loss": 0.6562,
+      "step": 3190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662730280213843e-05,
+      "loss": 0.5963,
+      "step": 3191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662441612854794e-05,
+      "loss": 0.5485,
+      "step": 3192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662152824134938e-05,
+      "loss": 0.5639,
+      "step": 3193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661863914057903e-05,
+      "loss": 0.6305,
+      "step": 3194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661574882627312e-05,
+      "loss": 0.6125,
+      "step": 3195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661285729846803e-05,
+      "loss": 0.6064,
+      "step": 3196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966099645572e-05,
+      "loss": 0.635,
+      "step": 3197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660707060250543e-05,
+      "loss": 0.5802,
+      "step": 3198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660417543442064e-05,
+      "loss": 0.6497,
+      "step": 3199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96601279052982e-05,
+      "loss": 0.5642,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659838145822592e-05,
+      "loss": 0.6367,
+      "step": 3201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965954826501887e-05,
+      "loss": 0.5739,
+      "step": 3202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.7151,
+      "step": 3203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658968139441673e-05,
+      "loss": 0.5307,
+      "step": 3204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658677894675483e-05,
+      "loss": 0.5761,
+      "step": 3205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965838752859575e-05,
+      "loss": 0.5676,
+      "step": 3206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658097041206136e-05,
+      "loss": 0.5401,
+      "step": 3207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657806432510283e-05,
+      "loss": 0.551,
+      "step": 3208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657515702511837e-05,
+      "loss": 0.5874,
+      "step": 3209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657224851214455e-05,
+      "loss": 0.5523,
+      "step": 3210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656933878621786e-05,
+      "loss": 0.6158,
+      "step": 3211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656642784737485e-05,
+      "loss": 0.5305,
+      "step": 3212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656351569565214e-05,
+      "loss": 0.654,
+      "step": 3213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656060233108624e-05,
+      "loss": 0.6111,
+      "step": 3214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655768775371376e-05,
+      "loss": 0.6071,
+      "step": 3215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655477196357134e-05,
+      "loss": 0.5979,
+      "step": 3216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655185496069555e-05,
+      "loss": 0.5717,
+      "step": 3217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654893674512307e-05,
+      "loss": 0.6379,
+      "step": 3218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654601731689055e-05,
+      "loss": 0.6449,
+      "step": 3219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965430966760346e-05,
+      "loss": 0.5557,
+      "step": 3220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654017482259196e-05,
+      "loss": 0.6316,
+      "step": 3221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653725175659935e-05,
+      "loss": 0.599,
+      "step": 3222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653432747809344e-05,
+      "loss": 0.5617,
+      "step": 3223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653140198711096e-05,
+      "loss": 0.5567,
+      "step": 3224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652847528368867e-05,
+      "loss": 0.5646,
+      "step": 3225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652554736786334e-05,
+      "loss": 0.62,
+      "step": 3226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652261823967168e-05,
+      "loss": 0.5718,
+      "step": 3227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965196878991506e-05,
+      "loss": 0.5515,
+      "step": 3228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651675634633678e-05,
+      "loss": 0.636,
+      "step": 3229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651382358126713e-05,
+      "loss": 0.6315,
+      "step": 3230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965108896039784e-05,
+      "loss": 0.642,
+      "step": 3231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650795441450758e-05,
+      "loss": 0.5519,
+      "step": 3232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650501801289133e-05,
+      "loss": 0.6028,
+      "step": 3233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650208039916674e-05,
+      "loss": 0.5823,
+      "step": 3234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964991415733706e-05,
+      "loss": 0.6151,
+      "step": 3235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649620153553982e-05,
+      "loss": 0.5828,
+      "step": 3236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649326028571136e-05,
+      "loss": 0.5731,
+      "step": 3237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964903178239221e-05,
+      "loss": 0.608,
+      "step": 3238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648737415020908e-05,
+      "loss": 0.5818,
+      "step": 3239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648442926460926e-05,
+      "loss": 0.5321,
+      "step": 3240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648148316715957e-05,
+      "loss": 0.6131,
+      "step": 3241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647853585789704e-05,
+      "loss": 0.5924,
+      "step": 3242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964755873368587e-05,
+      "loss": 0.5931,
+      "step": 3243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964726376040816e-05,
+      "loss": 0.6147,
+      "step": 3244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646968665960277e-05,
+      "loss": 0.5902,
+      "step": 3245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646673450345926e-05,
+      "loss": 0.5853,
+      "step": 3246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646378113568818e-05,
+      "loss": 0.6184,
+      "step": 3247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964608265563266e-05,
+      "loss": 0.6683,
+      "step": 3248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645787076541162e-05,
+      "loss": 0.6373,
+      "step": 3249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964549137629804e-05,
+      "loss": 0.6672,
+      "step": 3250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645195554907004e-05,
+      "loss": 0.6406,
+      "step": 3251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644899612371773e-05,
+      "loss": 0.6273,
+      "step": 3252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964460354869606e-05,
+      "loss": 0.6124,
+      "step": 3253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964430736388359e-05,
+      "loss": 0.5885,
+      "step": 3254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964401105793808e-05,
+      "loss": 0.5593,
+      "step": 3255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964371463086325e-05,
+      "loss": 0.5709,
+      "step": 3256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964341808266282e-05,
+      "loss": 0.6114,
+      "step": 3257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643121413340524e-05,
+      "loss": 0.6625,
+      "step": 3258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964282462290008e-05,
+      "loss": 0.6202,
+      "step": 3259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964252771134522e-05,
+      "loss": 0.5625,
+      "step": 3260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964223067867967e-05,
+      "loss": 0.596,
+      "step": 3261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641933524907167e-05,
+      "loss": 0.593,
+      "step": 3262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641636250031434e-05,
+      "loss": 0.6261,
+      "step": 3263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641338854056214e-05,
+      "loss": 0.6529,
+      "step": 3264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641041336985234e-05,
+      "loss": 0.5672,
+      "step": 3265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640743698822237e-05,
+      "loss": 0.6881,
+      "step": 3266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964044593957096e-05,
+      "loss": 0.6159,
+      "step": 3267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964014805923514e-05,
+      "loss": 0.6346,
+      "step": 3268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963985005781852e-05,
+      "loss": 0.585,
+      "step": 3269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639551935324846e-05,
+      "loss": 0.5983,
+      "step": 3270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639253691757857e-05,
+      "loss": 0.6203,
+      "step": 3271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638955327121305e-05,
+      "loss": 0.5956,
+      "step": 3272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638656841418933e-05,
+      "loss": 0.5714,
+      "step": 3273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963835823465449e-05,
+      "loss": 0.6121,
+      "step": 3274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963805950683173e-05,
+      "loss": 0.5265,
+      "step": 3275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96377606579544e-05,
+      "loss": 0.5587,
+      "step": 3276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963746168802626e-05,
+      "loss": 0.5391,
+      "step": 3277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963716259705106e-05,
+      "loss": 0.6185,
+      "step": 3278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636863385032555e-05,
+      "loss": 0.6448,
+      "step": 3279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636564051974507e-05,
+      "loss": 0.6463,
+      "step": 3280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636264597880678e-05,
+      "loss": 0.6245,
+      "step": 3281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635965022754822e-05,
+      "loss": 0.5931,
+      "step": 3282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963566532660071e-05,
+      "loss": 0.6282,
+      "step": 3283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96353655094221e-05,
+      "loss": 0.5406,
+      "step": 3284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963506557122276e-05,
+      "loss": 0.6273,
+      "step": 3285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634765512006454e-05,
+      "loss": 0.6338,
+      "step": 3286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634465331776958e-05,
+      "loss": 0.6265,
+      "step": 3287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963416503053803e-05,
+      "loss": 0.5907,
+      "step": 3288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963386460829346e-05,
+      "loss": 0.6116,
+      "step": 3289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633564065047003e-05,
+      "loss": 0.6265,
+      "step": 3290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633263400802448e-05,
+      "loss": 0.5814,
+      "step": 3291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963296261556356e-05,
+      "loss": 0.6033,
+      "step": 3292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632661709334124e-05,
+      "loss": 0.6198,
+      "step": 3293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963236068211792e-05,
+      "loss": 0.6099,
+      "step": 3294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963205953391872e-05,
+      "loss": 0.5994,
+      "step": 3295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631758264740316e-05,
+      "loss": 0.5587,
+      "step": 3296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963145687458649e-05,
+      "loss": 0.6058,
+      "step": 3297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631155363461027e-05,
+      "loss": 0.6139,
+      "step": 3298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630853731367715e-05,
+      "loss": 0.5444,
+      "step": 3299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630551978310336e-05,
+      "loss": 0.5776,
+      "step": 3300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630250104292684e-05,
+      "loss": 0.6727,
+      "step": 3301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629948109318554e-05,
+      "loss": 0.5282,
+      "step": 3302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962964599339174e-05,
+      "loss": 0.612,
+      "step": 3303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629343756516024e-05,
+      "loss": 0.5211,
+      "step": 3304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962904139869522e-05,
+      "loss": 0.5825,
+      "step": 3305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628738919933114e-05,
+      "loss": 0.6895,
+      "step": 3306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628436320233506e-05,
+      "loss": 0.5732,
+      "step": 3307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96281335996002e-05,
+      "loss": 0.5851,
+      "step": 3308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627830758036995e-05,
+      "loss": 0.5821,
+      "step": 3309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627527795547698e-05,
+      "loss": 0.5889,
+      "step": 3310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627224712136113e-05,
+      "loss": 0.6026,
+      "step": 3311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626921507806044e-05,
+      "loss": 0.5748,
+      "step": 3312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626618182561305e-05,
+      "loss": 0.6271,
+      "step": 3313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96263147364057e-05,
+      "loss": 0.5974,
+      "step": 3314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626011169343043e-05,
+      "loss": 0.6161,
+      "step": 3315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962570748137715e-05,
+      "loss": 0.5821,
+      "step": 3316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625403672511826e-05,
+      "loss": 0.5928,
+      "step": 3317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625099742750896e-05,
+      "loss": 0.5885,
+      "step": 3318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962479569209818e-05,
+      "loss": 0.6834,
+      "step": 3319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962449152055748e-05,
+      "loss": 0.6169,
+      "step": 3320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624187228132633e-05,
+      "loss": 0.6096,
+      "step": 3321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623882814827455e-05,
+      "loss": 0.5792,
+      "step": 3322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962357828064577e-05,
+      "loss": 0.5723,
+      "step": 3323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623273625591404e-05,
+      "loss": 0.5494,
+      "step": 3324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962296884966818e-05,
+      "loss": 0.6188,
+      "step": 3325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962266395287993e-05,
+      "loss": 0.6411,
+      "step": 3326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622358935230482e-05,
+      "loss": 0.5994,
+      "step": 3327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622053796723667e-05,
+      "loss": 0.6555,
+      "step": 3328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621748537363316e-05,
+      "loss": 0.599,
+      "step": 3329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621443157153264e-05,
+      "loss": 0.755,
+      "step": 3330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621137656097347e-05,
+      "loss": 0.651,
+      "step": 3331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620832034199404e-05,
+      "loss": 0.604,
+      "step": 3332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620526291463272e-05,
+      "loss": 0.5681,
+      "step": 3333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962022042789279e-05,
+      "loss": 0.5936,
+      "step": 3334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96199144434918e-05,
+      "loss": 0.6248,
+      "step": 3335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619608338264145e-05,
+      "loss": 0.6589,
+      "step": 3336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619302112213674e-05,
+      "loss": 0.5536,
+      "step": 3337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961899576534423e-05,
+      "loss": 0.5514,
+      "step": 3338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618689297659655e-05,
+      "loss": 0.5349,
+      "step": 3339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961838270916381e-05,
+      "loss": 0.6051,
+      "step": 3340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618075999860536e-05,
+      "loss": 0.4542,
+      "step": 3341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961776916975369e-05,
+      "loss": 0.5112,
+      "step": 3342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617462218847122e-05,
+      "loss": 0.6475,
+      "step": 3343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617155147144693e-05,
+      "loss": 0.6064,
+      "step": 3344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616847954650255e-05,
+      "loss": 0.6167,
+      "step": 3345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616540641367666e-05,
+      "loss": 0.6631,
+      "step": 3346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961623320730079e-05,
+      "loss": 0.5855,
+      "step": 3347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961592565245349e-05,
+      "loss": 0.6587,
+      "step": 3348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615617976829622e-05,
+      "loss": 0.566,
+      "step": 3349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961531018043305e-05,
+      "loss": 0.5422,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615002263267652e-05,
+      "loss": 0.5438,
+      "step": 3351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961469422533728e-05,
+      "loss": 0.5928,
+      "step": 3352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961438606664581e-05,
+      "loss": 0.5635,
+      "step": 3353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961407778719712e-05,
+      "loss": 0.5031,
+      "step": 3354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613769386995066e-05,
+      "loss": 0.5613,
+      "step": 3355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613460866043533e-05,
+      "loss": 0.5748,
+      "step": 3356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613152224346396e-05,
+      "loss": 0.5854,
+      "step": 3357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612843461907528e-05,
+      "loss": 0.5775,
+      "step": 3358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612534578730808e-05,
+      "loss": 0.5884,
+      "step": 3359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961222557482011e-05,
+      "loss": 0.6189,
+      "step": 3360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611916450179324e-05,
+      "loss": 0.5793,
+      "step": 3361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961160720481233e-05,
+      "loss": 0.5595,
+      "step": 3362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961129783872301e-05,
+      "loss": 0.6328,
+      "step": 3363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961098835191525e-05,
+      "loss": 0.5794,
+      "step": 3364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961067874439294e-05,
+      "loss": 0.6036,
+      "step": 3365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610369016159965e-05,
+      "loss": 0.5882,
+      "step": 3366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610059167220216e-05,
+      "loss": 0.597,
+      "step": 3367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609749197577588e-05,
+      "loss": 0.6003,
+      "step": 3368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609439107235973e-05,
+      "loss": 0.5507,
+      "step": 3369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609128896199264e-05,
+      "loss": 0.6524,
+      "step": 3370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608818564471355e-05,
+      "loss": 0.559,
+      "step": 3371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960850811205615e-05,
+      "loss": 0.5606,
+      "step": 3372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608197538957542e-05,
+      "loss": 0.6745,
+      "step": 3373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607886845179437e-05,
+      "loss": 0.6519,
+      "step": 3374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607576030725733e-05,
+      "loss": 0.6022,
+      "step": 3375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607265095600335e-05,
+      "loss": 0.6337,
+      "step": 3376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606954039807155e-05,
+      "loss": 0.6319,
+      "step": 3377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960664286335009e-05,
+      "loss": 0.5765,
+      "step": 3378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606331566233052e-05,
+      "loss": 0.584,
+      "step": 3379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606020148459956e-05,
+      "loss": 0.6442,
+      "step": 3380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605708610034704e-05,
+      "loss": 0.6651,
+      "step": 3381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605396950961216e-05,
+      "loss": 0.5261,
+      "step": 3382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605085171243403e-05,
+      "loss": 0.5821,
+      "step": 3383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604773270885184e-05,
+      "loss": 0.5571,
+      "step": 3384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604461249890473e-05,
+      "loss": 0.5775,
+      "step": 3385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604149108263193e-05,
+      "loss": 0.576,
+      "step": 3386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603836846007264e-05,
+      "loss": 0.5736,
+      "step": 3387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603524463126603e-05,
+      "loss": 0.63,
+      "step": 3388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603211959625136e-05,
+      "loss": 0.5467,
+      "step": 3389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960289933550679e-05,
+      "loss": 0.5791,
+      "step": 3390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602586590775492e-05,
+      "loss": 0.5731,
+      "step": 3391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602273725435167e-05,
+      "loss": 0.5163,
+      "step": 3392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601960739489745e-05,
+      "loss": 0.552,
+      "step": 3393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960164763294316e-05,
+      "loss": 0.6119,
+      "step": 3394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601334405799344e-05,
+      "loss": 0.558,
+      "step": 3395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960102105806223e-05,
+      "loss": 0.6157,
+      "step": 3396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600707589735752e-05,
+      "loss": 0.5635,
+      "step": 3397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960039400082385e-05,
+      "loss": 0.6291,
+      "step": 3398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960008029133046e-05,
+      "loss": 0.556,
+      "step": 3399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599766461259525e-05,
+      "loss": 0.6379,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599452510614987e-05,
+      "loss": 0.6264,
+      "step": 3401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959913843940079e-05,
+      "loss": 0.5692,
+      "step": 3402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598824247620875e-05,
+      "loss": 0.5664,
+      "step": 3403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598509935279188e-05,
+      "loss": 0.6657,
+      "step": 3404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598195502379677e-05,
+      "loss": 0.5314,
+      "step": 3405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597880948926294e-05,
+      "loss": 0.6257,
+      "step": 3406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959756627492299e-05,
+      "loss": 0.6087,
+      "step": 3407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597251480373715e-05,
+      "loss": 0.5803,
+      "step": 3408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596936565282427e-05,
+      "loss": 0.5096,
+      "step": 3409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596621529653075e-05,
+      "loss": 0.5824,
+      "step": 3410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959630637348962e-05,
+      "loss": 0.5868,
+      "step": 3411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959599109679602e-05,
+      "loss": 0.5696,
+      "step": 3412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595675699576232e-05,
+      "loss": 0.5977,
+      "step": 3413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595360181834226e-05,
+      "loss": 0.6276,
+      "step": 3414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595044543573953e-05,
+      "loss": 0.6453,
+      "step": 3415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959472878479938e-05,
+      "loss": 0.5708,
+      "step": 3416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594412905514483e-05,
+      "loss": 0.6551,
+      "step": 3417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959409690572322e-05,
+      "loss": 0.5971,
+      "step": 3418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9593780785429563e-05,
+      "loss": 0.5704,
+      "step": 3419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593464544637482e-05,
+      "loss": 0.5949,
+      "step": 3420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959314818335095e-05,
+      "loss": 0.6125,
+      "step": 3421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959283170157394e-05,
+      "loss": 0.5262,
+      "step": 3422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592515099310422e-05,
+      "loss": 0.6577,
+      "step": 3423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592198376564383e-05,
+      "loss": 0.5592,
+      "step": 3424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959188153333979e-05,
+      "loss": 0.6109,
+      "step": 3425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959156456964063e-05,
+      "loss": 0.5608,
+      "step": 3426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959124748547088e-05,
+      "loss": 0.5328,
+      "step": 3427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590930280834527e-05,
+      "loss": 0.5434,
+      "step": 3428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590612955735553e-05,
+      "loss": 0.5703,
+      "step": 3429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959029551017794e-05,
+      "loss": 0.6003,
+      "step": 3430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589977944165682e-05,
+      "loss": 0.5867,
+      "step": 3431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958966025770276e-05,
+      "loss": 0.5898,
+      "step": 3432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958934245079317e-05,
+      "loss": 0.6121,
+      "step": 3433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589024523440904e-05,
+      "loss": 0.5318,
+      "step": 3434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588706475649953e-05,
+      "loss": 0.6424,
+      "step": 3435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588388307424307e-05,
+      "loss": 0.5584,
+      "step": 3436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588070018767968e-05,
+      "loss": 0.6241,
+      "step": 3437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587751609684935e-05,
+      "loss": 0.6017,
+      "step": 3438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587433080179202e-05,
+      "loss": 0.5854,
+      "step": 3439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587114430254776e-05,
+      "loss": 0.5987,
+      "step": 3440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586795659915654e-05,
+      "loss": 0.5934,
+      "step": 3441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586476769165843e-05,
+      "loss": 0.5692,
+      "step": 3442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586157758009342e-05,
+      "loss": 0.6995,
+      "step": 3443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585838626450167e-05,
+      "loss": 0.5523,
+      "step": 3444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585519374492324e-05,
+      "loss": 0.5513,
+      "step": 3445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585200002139815e-05,
+      "loss": 0.6394,
+      "step": 3446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958488050939666e-05,
+      "loss": 0.5246,
+      "step": 3447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584560896266872e-05,
+      "loss": 0.5435,
+      "step": 3448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584241162754457e-05,
+      "loss": 0.5681,
+      "step": 3449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958392130886344e-05,
+      "loss": 0.6434,
+      "step": 3450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583601334597834e-05,
+      "loss": 0.5961,
+      "step": 3451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583281239961656e-05,
+      "loss": 0.567,
+      "step": 3452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582961024958932e-05,
+      "loss": 0.6278,
+      "step": 3453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958264068959368e-05,
+      "loss": 0.6213,
+      "step": 3454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582320233869922e-05,
+      "loss": 0.6011,
+      "step": 3455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958199965779169e-05,
+      "loss": 0.5754,
+      "step": 3456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581678961363e-05,
+      "loss": 0.6029,
+      "step": 3457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958135814458789e-05,
+      "loss": 0.598,
+      "step": 3458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581037207470382e-05,
+      "loss": 0.5323,
+      "step": 3459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580716150014513e-05,
+      "loss": 0.6255,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580394972224312e-05,
+      "loss": 0.5878,
+      "step": 3461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958007367410381e-05,
+      "loss": 0.5923,
+      "step": 3462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579752255657048e-05,
+      "loss": 0.5607,
+      "step": 3463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579430716888063e-05,
+      "loss": 0.5681,
+      "step": 3464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579109057800893e-05,
+      "loss": 0.5046,
+      "step": 3465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578787278399575e-05,
+      "loss": 0.5646,
+      "step": 3466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957846537868815e-05,
+      "loss": 0.5778,
+      "step": 3467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578143358670663e-05,
+      "loss": 0.5425,
+      "step": 3468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577821218351163e-05,
+      "loss": 0.5646,
+      "step": 3469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577498957733687e-05,
+      "loss": 0.608,
+      "step": 3470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577176576822292e-05,
+      "loss": 0.5657,
+      "step": 3471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957685407562102e-05,
+      "loss": 0.6383,
+      "step": 3472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576531454133926e-05,
+      "loss": 0.5739,
+      "step": 3473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576208712365062e-05,
+      "loss": 0.6083,
+      "step": 3474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957588585031848e-05,
+      "loss": 0.5558,
+      "step": 3475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575562867998235e-05,
+      "loss": 0.5664,
+      "step": 3476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575239765408382e-05,
+      "loss": 0.5973,
+      "step": 3477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574916542552983e-05,
+      "loss": 0.6103,
+      "step": 3478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574593199436098e-05,
+      "loss": 0.5299,
+      "step": 3479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957426973606178e-05,
+      "loss": 0.5597,
+      "step": 3480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573946152434104e-05,
+      "loss": 0.6315,
+      "step": 3481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573622448557124e-05,
+      "loss": 0.5439,
+      "step": 3482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957329862443491e-05,
+      "loss": 0.6258,
+      "step": 3483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957297468007153e-05,
+      "loss": 0.6239,
+      "step": 3484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572650615471053e-05,
+      "loss": 0.5669,
+      "step": 3485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957232643063755e-05,
+      "loss": 0.6193,
+      "step": 3486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572002125575088e-05,
+      "loss": 0.6305,
+      "step": 3487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571677700287743e-05,
+      "loss": 0.6967,
+      "step": 3488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571353154779593e-05,
+      "loss": 0.637,
+      "step": 3489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957102848905471e-05,
+      "loss": 0.6301,
+      "step": 3490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957070370311717e-05,
+      "loss": 0.5699,
+      "step": 3491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957037879697106e-05,
+      "loss": 0.5914,
+      "step": 3492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957005377062045e-05,
+      "loss": 0.57,
+      "step": 3493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569728624069433e-05,
+      "loss": 0.6163,
+      "step": 3494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.626,
+      "step": 3495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569077970382506e-05,
+      "loss": 0.5868,
+      "step": 3496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956875246325476e-05,
+      "loss": 0.6596,
+      "step": 3497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568426835942953e-05,
+      "loss": 0.5963,
+      "step": 3498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568101088451164e-05,
+      "loss": 0.5596,
+      "step": 3499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567775220783492e-05,
+      "loss": 0.571,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956744923294403e-05,
+      "loss": 0.6679,
+      "step": 3501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956712312493687e-05,
+      "loss": 0.5796,
+      "step": 3502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566796896766103e-05,
+      "loss": 0.5461,
+      "step": 3503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566470548435833e-05,
+      "loss": 0.6456,
+      "step": 3504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956614407995016e-05,
+      "loss": 0.6007,
+      "step": 3505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956581749131318e-05,
+      "loss": 0.6251,
+      "step": 3506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565490782528997e-05,
+      "loss": 0.5939,
+      "step": 3507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565163953601712e-05,
+      "loss": 0.6251,
+      "step": 3508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564837004535432e-05,
+      "loss": 0.6573,
+      "step": 3509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564509935334265e-05,
+      "loss": 0.6201,
+      "step": 3510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956418274600232e-05,
+      "loss": 0.5359,
+      "step": 3511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563855436543696e-05,
+      "loss": 0.5989,
+      "step": 3512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956352800696252e-05,
+      "loss": 0.5767,
+      "step": 3513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563200457262896e-05,
+      "loss": 0.5674,
+      "step": 3514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956287278744893e-05,
+      "loss": 0.6037,
+      "step": 3515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9562544997524754e-05,
+      "loss": 0.5939,
+      "step": 3516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956221708749448e-05,
+      "loss": 0.5509,
+      "step": 3517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561889057362216e-05,
+      "loss": 0.559,
+      "step": 3518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561560907132096e-05,
+      "loss": 0.5782,
+      "step": 3519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561232636808232e-05,
+      "loss": 0.6337,
+      "step": 3520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560904246394754e-05,
+      "loss": 0.6643,
+      "step": 3521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560575735895782e-05,
+      "loss": 0.6067,
+      "step": 3522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560247105315443e-05,
+      "loss": 0.5973,
+      "step": 3523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955991835465786e-05,
+      "loss": 0.5975,
+      "step": 3524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559589483927173e-05,
+      "loss": 0.6182,
+      "step": 3525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559260493127503e-05,
+      "loss": 0.6483,
+      "step": 3526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558931382262992e-05,
+      "loss": 0.5034,
+      "step": 3527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558602151337762e-05,
+      "loss": 0.632,
+      "step": 3528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558272800355955e-05,
+      "loss": 0.7017,
+      "step": 3529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557943329321706e-05,
+      "loss": 0.6329,
+      "step": 3530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557613738239154e-05,
+      "loss": 0.6626,
+      "step": 3531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557284027112437e-05,
+      "loss": 0.5696,
+      "step": 3532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556954195945697e-05,
+      "loss": 0.6311,
+      "step": 3533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955662424474308e-05,
+      "loss": 0.5594,
+      "step": 3534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556294173508724e-05,
+      "loss": 0.6674,
+      "step": 3535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955596398224678e-05,
+      "loss": 0.5975,
+      "step": 3536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955563367096139e-05,
+      "loss": 0.6545,
+      "step": 3537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555303239656714e-05,
+      "loss": 0.5969,
+      "step": 3538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554972688336888e-05,
+      "loss": 0.6064,
+      "step": 3539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955464201700607e-05,
+      "loss": 0.5983,
+      "step": 3540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554311225668412e-05,
+      "loss": 0.6126,
+      "step": 3541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553980314328073e-05,
+      "loss": 0.5647,
+      "step": 3542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553649282989204e-05,
+      "loss": 0.5648,
+      "step": 3543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553318131655965e-05,
+      "loss": 0.5956,
+      "step": 3544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552986860332515e-05,
+      "loss": 0.5871,
+      "step": 3545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552655469023017e-05,
+      "loss": 0.6375,
+      "step": 3546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955232395773163e-05,
+      "loss": 0.5199,
+      "step": 3547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955199232646252e-05,
+      "loss": 0.5665,
+      "step": 3548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551660575219853e-05,
+      "loss": 0.628,
+      "step": 3549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955132870400779e-05,
+      "loss": 0.5799,
+      "step": 3550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955099671283051e-05,
+      "loss": 0.5386,
+      "step": 3551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955066460169217e-05,
+      "loss": 0.6304,
+      "step": 3552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955033237059695e-05,
+      "loss": 0.5647,
+      "step": 3553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550000019549016e-05,
+      "loss": 0.6189,
+      "step": 3554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549667548552557e-05,
+      "loss": 0.577,
+      "step": 3555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954933495761173e-05,
+      "loss": 0.6194,
+      "step": 3556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549002246730724e-05,
+      "loss": 0.556,
+      "step": 3557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548669415913716e-05,
+      "loss": 0.568,
+      "step": 3558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548336465164883e-05,
+      "loss": 0.5933,
+      "step": 3559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548003394488412e-05,
+      "loss": 0.6011,
+      "step": 3560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954767020388848e-05,
+      "loss": 0.6153,
+      "step": 3561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954733689336928e-05,
+      "loss": 0.6149,
+      "step": 3562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954700346293499e-05,
+      "loss": 0.5513,
+      "step": 3563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546669912589803e-05,
+      "loss": 0.5246,
+      "step": 3564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954633624233791e-05,
+      "loss": 0.5895,
+      "step": 3565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546002452183495e-05,
+      "loss": 0.5992,
+      "step": 3566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545668542130757e-05,
+      "loss": 0.5466,
+      "step": 3567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545334512183886e-05,
+      "loss": 0.5698,
+      "step": 3568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545000362347078e-05,
+      "loss": 0.5399,
+      "step": 3569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544666092624537e-05,
+      "loss": 0.5502,
+      "step": 3570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954433170302045e-05,
+      "loss": 0.5384,
+      "step": 3571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543997193539022e-05,
+      "loss": 0.6492,
+      "step": 3572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543662564184455e-05,
+      "loss": 0.6436,
+      "step": 3573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543327814960953e-05,
+      "loss": 0.6109,
+      "step": 3574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542992945872718e-05,
+      "loss": 0.5626,
+      "step": 3575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954265795692396e-05,
+      "loss": 0.5321,
+      "step": 3576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542322848118882e-05,
+      "loss": 0.5667,
+      "step": 3577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541987619461694e-05,
+      "loss": 0.5381,
+      "step": 3578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541652270956605e-05,
+      "loss": 0.5794,
+      "step": 3579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541316802607834e-05,
+      "loss": 0.5823,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540981214419587e-05,
+      "loss": 0.5604,
+      "step": 3581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954064550639608e-05,
+      "loss": 0.5539,
+      "step": 3582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540309678541534e-05,
+      "loss": 0.6573,
+      "step": 3583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953997373086016e-05,
+      "loss": 0.4914,
+      "step": 3584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539637663356187e-05,
+      "loss": 0.633,
+      "step": 3585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539301476033828e-05,
+      "loss": 0.69,
+      "step": 3586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538965168897307e-05,
+      "loss": 0.5481,
+      "step": 3587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953862874195085e-05,
+      "loss": 0.5752,
+      "step": 3588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538292195198682e-05,
+      "loss": 0.5441,
+      "step": 3589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953795552864503e-05,
+      "loss": 0.59,
+      "step": 3590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953761874229412e-05,
+      "loss": 0.5769,
+      "step": 3591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537281836150188e-05,
+      "loss": 0.5953,
+      "step": 3592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536944810217463e-05,
+      "loss": 0.6313,
+      "step": 3593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536607664500177e-05,
+      "loss": 0.5811,
+      "step": 3594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953627039900256e-05,
+      "loss": 0.5468,
+      "step": 3595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535933013728856e-05,
+      "loss": 0.6382,
+      "step": 3596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535595508683295e-05,
+      "loss": 0.5941,
+      "step": 3597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535257883870127e-05,
+      "loss": 0.5704,
+      "step": 3598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534920139293582e-05,
+      "loss": 0.6004,
+      "step": 3599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534582274957908e-05,
+      "loss": 0.5961,
+      "step": 3600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534244290867344e-05,
+      "loss": 0.591,
+      "step": 3601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533906187026138e-05,
+      "loss": 0.6728,
+      "step": 3602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953356796343854e-05,
+      "loss": 0.6203,
+      "step": 3603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533229620108793e-05,
+      "loss": 0.5805,
+      "step": 3604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532891157041144e-05,
+      "loss": 0.6055,
+      "step": 3605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532552574239853e-05,
+      "loss": 0.5501,
+      "step": 3606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532213871709163e-05,
+      "loss": 0.556,
+      "step": 3607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531875049453338e-05,
+      "loss": 0.6121,
+      "step": 3608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953153610747663e-05,
+      "loss": 0.6277,
+      "step": 3609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531197045783288e-05,
+      "loss": 0.643,
+      "step": 3610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953085786437758e-05,
+      "loss": 0.6432,
+      "step": 3611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9530518563263765e-05,
+      "loss": 0.5075,
+      "step": 3612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95301791424461e-05,
+      "loss": 0.6362,
+      "step": 3613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529839601928854e-05,
+      "loss": 0.6248,
+      "step": 3614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529499941716286e-05,
+      "loss": 0.5612,
+      "step": 3615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529160161812668e-05,
+      "loss": 0.6176,
+      "step": 3616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528820262222264e-05,
+      "loss": 0.6315,
+      "step": 3617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528480242949347e-05,
+      "loss": 0.6187,
+      "step": 3618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952814010399818e-05,
+      "loss": 0.632,
+      "step": 3619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527799845373037e-05,
+      "loss": 0.5319,
+      "step": 3620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95274594670782e-05,
+      "loss": 0.6404,
+      "step": 3621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527118969117938e-05,
+      "loss": 0.61,
+      "step": 3622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526778351496526e-05,
+      "loss": 0.6095,
+      "step": 3623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526437614218245e-05,
+      "loss": 0.6106,
+      "step": 3624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526096757287376e-05,
+      "loss": 0.5986,
+      "step": 3625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525755780708193e-05,
+      "loss": 0.6411,
+      "step": 3626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952541468448499e-05,
+      "loss": 0.6125,
+      "step": 3627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952507346862204e-05,
+      "loss": 0.6427,
+      "step": 3628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524732133123635e-05,
+      "loss": 0.5995,
+      "step": 3629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524390677994063e-05,
+      "loss": 0.5975,
+      "step": 3630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952404910323761e-05,
+      "loss": 0.5804,
+      "step": 3631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523707408858565e-05,
+      "loss": 0.5944,
+      "step": 3632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523365594861225e-05,
+      "loss": 0.6522,
+      "step": 3633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952302366124988e-05,
+      "loss": 0.6358,
+      "step": 3634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522681608028818e-05,
+      "loss": 0.5361,
+      "step": 3635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522339435202348e-05,
+      "loss": 0.5706,
+      "step": 3636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521997142774758e-05,
+      "loss": 0.6298,
+      "step": 3637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521654730750354e-05,
+      "loss": 0.5846,
+      "step": 3638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952131219913343e-05,
+      "loss": 0.7297,
+      "step": 3639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520969547928294e-05,
+      "loss": 0.5725,
+      "step": 3640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520626777139243e-05,
+      "loss": 0.6083,
+      "step": 3641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952028388677059e-05,
+      "loss": 0.5679,
+      "step": 3642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951994087682664e-05,
+      "loss": 0.6225,
+      "step": 3643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519597747311694e-05,
+      "loss": 0.6532,
+      "step": 3644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951925449823007e-05,
+      "loss": 0.5869,
+      "step": 3645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518911129586077e-05,
+      "loss": 0.6338,
+      "step": 3646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518567641384027e-05,
+      "loss": 0.5471,
+      "step": 3647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518224033628236e-05,
+      "loss": 0.5904,
+      "step": 3648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517880306323015e-05,
+      "loss": 0.5691,
+      "step": 3649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517536459472683e-05,
+      "loss": 0.5666,
+      "step": 3650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517192493081563e-05,
+      "loss": 0.5931,
+      "step": 3651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516848407153973e-05,
+      "loss": 0.5534,
+      "step": 3652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951650420169423e-05,
+      "loss": 0.6074,
+      "step": 3653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516159876706665e-05,
+      "loss": 0.585,
+      "step": 3654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515815432195598e-05,
+      "loss": 0.5938,
+      "step": 3655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515470868165356e-05,
+      "loss": 0.5308,
+      "step": 3656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515126184620267e-05,
+      "loss": 0.5517,
+      "step": 3657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514781381564663e-05,
+      "loss": 0.6475,
+      "step": 3658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514436459002867e-05,
+      "loss": 0.4876,
+      "step": 3659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951409141693922e-05,
+      "loss": 0.6464,
+      "step": 3660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951374625537805e-05,
+      "loss": 0.6157,
+      "step": 3661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513400974323695e-05,
+      "loss": 0.6526,
+      "step": 3662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951305557378049e-05,
+      "loss": 0.5815,
+      "step": 3663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512710053752776e-05,
+      "loss": 0.5553,
+      "step": 3664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951236441424489e-05,
+      "loss": 0.605,
+      "step": 3665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512018655261173e-05,
+      "loss": 0.545,
+      "step": 3666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951167277680597e-05,
+      "loss": 0.6526,
+      "step": 3667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511326778883622e-05,
+      "loss": 0.5183,
+      "step": 3668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951098066149848e-05,
+      "loss": 0.6186,
+      "step": 3669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510634424654888e-05,
+      "loss": 0.6228,
+      "step": 3670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510288068357194e-05,
+      "loss": 0.6104,
+      "step": 3671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950994159260975e-05,
+      "loss": 0.573,
+      "step": 3672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509594997416905e-05,
+      "loss": 0.6167,
+      "step": 3673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509248282783014e-05,
+      "loss": 0.7051,
+      "step": 3674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508901448712432e-05,
+      "loss": 0.6005,
+      "step": 3675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950855449520952e-05,
+      "loss": 0.5748,
+      "step": 3676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508207422278626e-05,
+      "loss": 0.6372,
+      "step": 3677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950786022992412e-05,
+      "loss": 0.5792,
+      "step": 3678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507512918150354e-05,
+      "loss": 0.5552,
+      "step": 3679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950716548696169e-05,
+      "loss": 0.5592,
+      "step": 3680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506817936362496e-05,
+      "loss": 0.6085,
+      "step": 3681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950647026635714e-05,
+      "loss": 0.5558,
+      "step": 3682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950612247694998e-05,
+      "loss": 0.6447,
+      "step": 3683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505774568145392e-05,
+      "loss": 0.6105,
+      "step": 3684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505426539947745e-05,
+      "loss": 0.6479,
+      "step": 3685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505078392361406e-05,
+      "loss": 0.672,
+      "step": 3686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504730125390748e-05,
+      "loss": 0.5386,
+      "step": 3687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504381739040154e-05,
+      "loss": 0.5686,
+      "step": 3688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504033233313987e-05,
+      "loss": 0.5778,
+      "step": 3689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503684608216633e-05,
+      "loss": 0.4988,
+      "step": 3690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950333586375247e-05,
+      "loss": 0.5422,
+      "step": 3691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950298699992587e-05,
+      "loss": 0.5919,
+      "step": 3692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502638016741226e-05,
+      "loss": 0.5447,
+      "step": 3693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502288914202913e-05,
+      "loss": 0.6274,
+      "step": 3694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501939692315322e-05,
+      "loss": 0.68,
+      "step": 3695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501590351082833e-05,
+      "loss": 0.5429,
+      "step": 3696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950124089050984e-05,
+      "loss": 0.5679,
+      "step": 3697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500891310600728e-05,
+      "loss": 0.6083,
+      "step": 3698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950054161135989e-05,
+      "loss": 0.6379,
+      "step": 3699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500191792791717e-05,
+      "loss": 0.6726,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499841854900604e-05,
+      "loss": 0.6212,
+      "step": 3701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499491797690944e-05,
+      "loss": 0.5999,
+      "step": 3702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499141621167136e-05,
+      "loss": 0.5752,
+      "step": 3703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498791325333574e-05,
+      "loss": 0.5317,
+      "step": 3704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498440910194665e-05,
+      "loss": 0.5681,
+      "step": 3705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498090375754805e-05,
+      "loss": 0.4997,
+      "step": 3706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497739722018397e-05,
+      "loss": 0.6132,
+      "step": 3707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497388948989844e-05,
+      "loss": 0.6007,
+      "step": 3708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497038056673555e-05,
+      "loss": 0.6226,
+      "step": 3709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949668704507394e-05,
+      "loss": 0.6053,
+      "step": 3710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94963359141954e-05,
+      "loss": 0.6027,
+      "step": 3711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495984664042352e-05,
+      "loss": 0.5357,
+      "step": 3712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94956332946192e-05,
+      "loss": 0.6262,
+      "step": 3713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949528180593037e-05,
+      "loss": 0.5,
+      "step": 3714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949493019798026e-05,
+      "loss": 0.6165,
+      "step": 3715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94945784707733e-05,
+      "loss": 0.6366,
+      "step": 3716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494226624313902e-05,
+      "loss": 0.602,
+      "step": 3717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493874658606486e-05,
+      "loss": 0.6306,
+      "step": 3718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949352257365547e-05,
+      "loss": 0.5794,
+      "step": 3719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493170369465285e-05,
+      "loss": 0.546,
+      "step": 3720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492818046040344e-05,
+      "loss": 0.6059,
+      "step": 3721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492465603385073e-05,
+      "loss": 0.5881,
+      "step": 3722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492113041503908e-05,
+      "loss": 0.6305,
+      "step": 3723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949176036040127e-05,
+      "loss": 0.5268,
+      "step": 3724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949140756008159e-05,
+      "loss": 0.5714,
+      "step": 3725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94910546405493e-05,
+      "loss": 0.6371,
+      "step": 3726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490701601808834e-05,
+      "loss": 0.6018,
+      "step": 3727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949034844386462e-05,
+      "loss": 0.5457,
+      "step": 3728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94899951667211e-05,
+      "loss": 0.5802,
+      "step": 3729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948964177038271e-05,
+      "loss": 0.5901,
+      "step": 3730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489288254853888e-05,
+      "loss": 0.5769,
+      "step": 3731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488934620139074e-05,
+      "loss": 0.6403,
+      "step": 3732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488580866242706e-05,
+      "loss": 0.6569,
+      "step": 3733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488226993169236e-05,
+      "loss": 0.576,
+      "step": 3734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94878730009231e-05,
+      "loss": 0.6039,
+      "step": 3735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948751888950875e-05,
+      "loss": 0.5995,
+      "step": 3736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948716465893063e-05,
+      "loss": 0.5433,
+      "step": 3737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948681030919319e-05,
+      "loss": 0.5533,
+      "step": 3738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486455840300883e-05,
+      "loss": 0.5968,
+      "step": 3739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486101252258157e-05,
+      "loss": 0.6162,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485746545069474e-05,
+      "loss": 0.5672,
+      "step": 3741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485391718739274e-05,
+      "loss": 0.5614,
+      "step": 3742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948503677327203e-05,
+      "loss": 0.6663,
+      "step": 3743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948468170867219e-05,
+      "loss": 0.5788,
+      "step": 3744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484326524944217e-05,
+      "loss": 0.6345,
+      "step": 3745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483971222092573e-05,
+      "loss": 0.5978,
+      "step": 3746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483615800121717e-05,
+      "loss": 0.6039,
+      "step": 3747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483260259036117e-05,
+      "loss": 0.605,
+      "step": 3748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482904598840236e-05,
+      "loss": 0.6006,
+      "step": 3749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482548819538542e-05,
+      "loss": 0.5834,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482192921135502e-05,
+      "loss": 0.6123,
+      "step": 3751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948183690363559e-05,
+      "loss": 0.571,
+      "step": 3752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481480767043277e-05,
+      "loss": 0.5788,
+      "step": 3753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481124511363034e-05,
+      "loss": 0.5864,
+      "step": 3754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480768136599337e-05,
+      "loss": 0.6372,
+      "step": 3755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480411642756658e-05,
+      "loss": 0.6464,
+      "step": 3756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948005502983948e-05,
+      "loss": 0.609,
+      "step": 3757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947969829785228e-05,
+      "loss": 0.5581,
+      "step": 3758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479341446799536e-05,
+      "loss": 0.6406,
+      "step": 3759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478984476685735e-05,
+      "loss": 0.6062,
+      "step": 3760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478627387515355e-05,
+      "loss": 0.6091,
+      "step": 3761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478270179292888e-05,
+      "loss": 0.5836,
+      "step": 3762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477912852022816e-05,
+      "loss": 0.5772,
+      "step": 3763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477555405709627e-05,
+      "loss": 0.5968,
+      "step": 3764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947719784035781e-05,
+      "loss": 0.5931,
+      "step": 3765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947684015597186e-05,
+      "loss": 0.6042,
+      "step": 3766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476482352556264e-05,
+      "loss": 0.5817,
+      "step": 3767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476124430115518e-05,
+      "loss": 0.65,
+      "step": 3768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475766388654123e-05,
+      "loss": 0.5374,
+      "step": 3769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947540822817657e-05,
+      "loss": 0.5901,
+      "step": 3770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947504994868736e-05,
+      "loss": 0.6239,
+      "step": 3771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474691550190987e-05,
+      "loss": 0.6517,
+      "step": 3772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947433303269196e-05,
+      "loss": 0.5732,
+      "step": 3773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947397439619478e-05,
+      "loss": 0.6444,
+      "step": 3774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473615640703955e-05,
+      "loss": 0.5737,
+      "step": 3775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947325676622398e-05,
+      "loss": 0.5194,
+      "step": 3776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472897772759372e-05,
+      "loss": 0.5957,
+      "step": 3777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947253866031464e-05,
+      "loss": 0.5872,
+      "step": 3778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472179428894288e-05,
+      "loss": 0.494,
+      "step": 3779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471820078502834e-05,
+      "loss": 0.6117,
+      "step": 3780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947146060914479e-05,
+      "loss": 0.5762,
+      "step": 3781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947110102082467e-05,
+      "loss": 0.6012,
+      "step": 3782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947074131354699e-05,
+      "loss": 0.506,
+      "step": 3783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470381487316266e-05,
+      "loss": 0.6049,
+      "step": 3784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470021542137027e-05,
+      "loss": 0.6014,
+      "step": 3785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946966147801378e-05,
+      "loss": 0.6053,
+      "step": 3786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.5733,
+      "step": 3787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468940992953382e-05,
+      "loss": 0.6376,
+      "step": 3788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468580572025277e-05,
+      "loss": 0.5817,
+      "step": 3789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468220032171266e-05,
+      "loss": 0.5234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467859373395882e-05,
+      "loss": 0.633,
+      "step": 3791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467498595703656e-05,
+      "loss": 0.5692,
+      "step": 3792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946713769909912e-05,
+      "loss": 0.6083,
+      "step": 3793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94667766835868e-05,
+      "loss": 0.5974,
+      "step": 3794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466415549171235e-05,
+      "loss": 0.6186,
+      "step": 3795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946605429585696e-05,
+      "loss": 0.6171,
+      "step": 3796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465692923648514e-05,
+      "loss": 0.578,
+      "step": 3797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465331432550435e-05,
+      "loss": 0.5634,
+      "step": 3798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464969822567267e-05,
+      "loss": 0.5728,
+      "step": 3799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946460809370354e-05,
+      "loss": 0.5277,
+      "step": 3800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946424624596381e-05,
+      "loss": 0.6433,
+      "step": 3801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463884279352616e-05,
+      "loss": 0.6183,
+      "step": 3802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463522193874506e-05,
+      "loss": 0.5877,
+      "step": 3803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463159989534024e-05,
+      "loss": 0.6313,
+      "step": 3804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946279766633573e-05,
+      "loss": 0.535,
+      "step": 3805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946243522428416e-05,
+      "loss": 0.6503,
+      "step": 3806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462072663383876e-05,
+      "loss": 0.6484,
+      "step": 3807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946170998363943e-05,
+      "loss": 0.6477,
+      "step": 3808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946134718505538e-05,
+      "loss": 0.5557,
+      "step": 3809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460984267636274e-05,
+      "loss": 0.584,
+      "step": 3810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946062123138668e-05,
+      "loss": 0.6273,
+      "step": 3811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946025807631115e-05,
+      "loss": 0.6124,
+      "step": 3812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945989480241425e-05,
+      "loss": 0.5667,
+      "step": 3813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459531409700543e-05,
+      "loss": 0.5917,
+      "step": 3814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459167898174586e-05,
+      "loss": 0.6246,
+      "step": 3815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458804267840957e-05,
+      "loss": 0.6199,
+      "step": 3816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458440518704216e-05,
+      "loss": 0.6737,
+      "step": 3817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945807665076893e-05,
+      "loss": 0.6169,
+      "step": 3818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457712664039673e-05,
+      "loss": 0.6178,
+      "step": 3819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457348558521012e-05,
+      "loss": 0.5607,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456984334217527e-05,
+      "loss": 0.5447,
+      "step": 3821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945661999113379e-05,
+      "loss": 0.5449,
+      "step": 3822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945625552927437e-05,
+      "loss": 0.5663,
+      "step": 3823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455890948643857e-05,
+      "loss": 0.606,
+      "step": 3824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455526249246822e-05,
+      "loss": 0.5513,
+      "step": 3825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455161431087847e-05,
+      "loss": 0.6273,
+      "step": 3826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454796494171515e-05,
+      "loss": 0.585,
+      "step": 3827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454431438502414e-05,
+      "loss": 0.6296,
+      "step": 3828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454066264085118e-05,
+      "loss": 0.577,
+      "step": 3829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453700970924224e-05,
+      "loss": 0.5084,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453335559024315e-05,
+      "loss": 0.5809,
+      "step": 3831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452970028389982e-05,
+      "loss": 0.628,
+      "step": 3832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452604379025816e-05,
+      "loss": 0.5423,
+      "step": 3833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452238610936412e-05,
+      "loss": 0.5759,
+      "step": 3834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945187272412636e-05,
+      "loss": 0.5985,
+      "step": 3835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451506718600253e-05,
+      "loss": 0.492,
+      "step": 3836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94511405943627e-05,
+      "loss": 0.5616,
+      "step": 3837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945077435141829e-05,
+      "loss": 0.6262,
+      "step": 3838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945040798977162e-05,
+      "loss": 0.6834,
+      "step": 3839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94500415094273e-05,
+      "loss": 0.6614,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449674910389928e-05,
+      "loss": 0.5836,
+      "step": 3841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944930819266411e-05,
+      "loss": 0.5144,
+      "step": 3842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448941356254453e-05,
+      "loss": 0.5727,
+      "step": 3843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448574401165562e-05,
+      "loss": 0.6215,
+      "step": 3844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944820732740205e-05,
+      "loss": 0.5977,
+      "step": 3845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944784013496852e-05,
+      "loss": 0.5401,
+      "step": 3846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447472823869594e-05,
+      "loss": 0.5204,
+      "step": 3847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447105394109878e-05,
+      "loss": 0.6522,
+      "step": 3848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446737845693987e-05,
+      "loss": 0.586,
+      "step": 3849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446370178626545e-05,
+      "loss": 0.6083,
+      "step": 3850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944600239291216e-05,
+      "loss": 0.5675,
+      "step": 3851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445634488555457e-05,
+      "loss": 0.5842,
+      "step": 3852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445266465561053e-05,
+      "loss": 0.5775,
+      "step": 3853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944489832393358e-05,
+      "loss": 0.5993,
+      "step": 3854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944453006367765e-05,
+      "loss": 0.5885,
+      "step": 3855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444161684797894e-05,
+      "loss": 0.5841,
+      "step": 3856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443793187298937e-05,
+      "loss": 0.5585,
+      "step": 3857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443424571185412e-05,
+      "loss": 0.5986,
+      "step": 3858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443055836461946e-05,
+      "loss": 0.6228,
+      "step": 3859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944268698313317e-05,
+      "loss": 0.5925,
+      "step": 3860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442318011203713e-05,
+      "loss": 0.6149,
+      "step": 3861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441948920678212e-05,
+      "loss": 0.5787,
+      "step": 3862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441579711561306e-05,
+      "loss": 0.621,
+      "step": 3863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441210383857632e-05,
+      "loss": 0.6325,
+      "step": 3864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440840937571823e-05,
+      "loss": 0.5952,
+      "step": 3865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440471372708526e-05,
+      "loss": 0.5568,
+      "step": 3866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944010168927238e-05,
+      "loss": 0.5865,
+      "step": 3867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439731887268026e-05,
+      "loss": 0.6005,
+      "step": 3868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943936196670011e-05,
+      "loss": 0.5735,
+      "step": 3869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438991927573282e-05,
+      "loss": 0.613,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438621769892188e-05,
+      "loss": 0.6167,
+      "step": 3871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438251493661473e-05,
+      "loss": 0.591,
+      "step": 3872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437881098885788e-05,
+      "loss": 0.6299,
+      "step": 3873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437510585569793e-05,
+      "loss": 0.5873,
+      "step": 3874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437139953718137e-05,
+      "loss": 0.5886,
+      "step": 3875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436769203335472e-05,
+      "loss": 0.4888,
+      "step": 3876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436398334426457e-05,
+      "loss": 0.6304,
+      "step": 3877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943602734699575e-05,
+      "loss": 0.5737,
+      "step": 3878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435656241048013e-05,
+      "loss": 0.6309,
+      "step": 3879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435285016587903e-05,
+      "loss": 0.5997,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434913673620087e-05,
+      "loss": 0.5912,
+      "step": 3881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434542212149227e-05,
+      "loss": 0.5944,
+      "step": 3882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434170632179986e-05,
+      "loss": 0.5924,
+      "step": 3883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433798933717033e-05,
+      "loss": 0.5062,
+      "step": 3884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943342711676504e-05,
+      "loss": 0.5426,
+      "step": 3885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433055181328672e-05,
+      "loss": 0.6017,
+      "step": 3886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432683127412605e-05,
+      "loss": 0.5257,
+      "step": 3887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432310955021507e-05,
+      "loss": 0.6311,
+      "step": 3888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431938664160055e-05,
+      "loss": 0.6085,
+      "step": 3889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431566254832924e-05,
+      "loss": 0.6237,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431193727044796e-05,
+      "loss": 0.5457,
+      "step": 3891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430821080800348e-05,
+      "loss": 0.5861,
+      "step": 3892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430448316104257e-05,
+      "loss": 0.6084,
+      "step": 3893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430075432961203e-05,
+      "loss": 0.5711,
+      "step": 3894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942970243137588e-05,
+      "loss": 0.5882,
+      "step": 3895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942932931135296e-05,
+      "loss": 0.6434,
+      "step": 3896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942895607289714e-05,
+      "loss": 0.6155,
+      "step": 3897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94285827160131e-05,
+      "loss": 0.6096,
+      "step": 3898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428209240705536e-05,
+      "loss": 0.569,
+      "step": 3899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942783564697914e-05,
+      "loss": 0.569,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9427461934838595e-05,
+      "loss": 0.6575,
+      "step": 3901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94270881042886e-05,
+      "loss": 0.6341,
+      "step": 3902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426714155333852e-05,
+      "loss": 0.6592,
+      "step": 3903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426340087979046e-05,
+      "loss": 0.5109,
+      "step": 3904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942596590222888e-05,
+      "loss": 0.5175,
+      "step": 3905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425591598088054e-05,
+      "loss": 0.5594,
+      "step": 3906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425217175561268e-05,
+      "loss": 0.663,
+      "step": 3907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424842634653228e-05,
+      "loss": 0.5177,
+      "step": 3908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942446797536864e-05,
+      "loss": 0.6055,
+      "step": 3909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424093197712205e-05,
+      "loss": 0.5772,
+      "step": 3910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423718301688627e-05,
+      "loss": 0.6034,
+      "step": 3911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423343287302624e-05,
+      "loss": 0.5868,
+      "step": 3912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422968154558897e-05,
+      "loss": 0.646,
+      "step": 3913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942259290346217e-05,
+      "loss": 0.6174,
+      "step": 3914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942221753401714e-05,
+      "loss": 0.5825,
+      "step": 3915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421842046228537e-05,
+      "loss": 0.5731,
+      "step": 3916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421466440101068e-05,
+      "loss": 0.5417,
+      "step": 3917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942109071563945e-05,
+      "loss": 0.5958,
+      "step": 3918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420714872848408e-05,
+      "loss": 0.6121,
+      "step": 3919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420338911732656e-05,
+      "loss": 0.6119,
+      "step": 3920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419962832296925e-05,
+      "loss": 0.5234,
+      "step": 3921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419586634545928e-05,
+      "loss": 0.5663,
+      "step": 3922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419210318484398e-05,
+      "loss": 0.5927,
+      "step": 3923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418833884117057e-05,
+      "loss": 0.542,
+      "step": 3924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941845733144864e-05,
+      "loss": 0.5331,
+      "step": 3925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418080660483866e-05,
+      "loss": 0.5395,
+      "step": 3926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417703871227475e-05,
+      "loss": 0.561,
+      "step": 3927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417326963684193e-05,
+      "loss": 0.5632,
+      "step": 3928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416949937858756e-05,
+      "loss": 0.6067,
+      "step": 3929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416572793755906e-05,
+      "loss": 0.5707,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416195531380372e-05,
+      "loss": 0.5933,
+      "step": 3931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415818150736893e-05,
+      "loss": 0.5919,
+      "step": 3932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941544065183021e-05,
+      "loss": 0.5741,
+      "step": 3933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415063034665066e-05,
+      "loss": 0.6299,
+      "step": 3934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414685299246204e-05,
+      "loss": 0.5486,
+      "step": 3935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414307445578366e-05,
+      "loss": 0.6031,
+      "step": 3936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94139294736663e-05,
+      "loss": 0.5549,
+      "step": 3937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413551383514752e-05,
+      "loss": 0.646,
+      "step": 3938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413173175128472e-05,
+      "loss": 0.5635,
+      "step": 3939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412794848512207e-05,
+      "loss": 0.5213,
+      "step": 3940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412416403670716e-05,
+      "loss": 0.6143,
+      "step": 3941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941203784060875e-05,
+      "loss": 0.5926,
+      "step": 3942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941165915933105e-05,
+      "loss": 0.6447,
+      "step": 3943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94112803598424e-05,
+      "loss": 0.5994,
+      "step": 3944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941090144214753e-05,
+      "loss": 0.6021,
+      "step": 3945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410522406251215e-05,
+      "loss": 0.5154,
+      "step": 3946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410143252158205e-05,
+      "loss": 0.6059,
+      "step": 3947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409763979873277e-05,
+      "loss": 0.5173,
+      "step": 3948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409384589401182e-05,
+      "loss": 0.6315,
+      "step": 3949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409005080746692e-05,
+      "loss": 0.5899,
+      "step": 3950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408625453914573e-05,
+      "loss": 0.6257,
+      "step": 3951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408245708909586e-05,
+      "loss": 0.6574,
+      "step": 3952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407865845736508e-05,
+      "loss": 0.574,
+      "step": 3953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940748586440011e-05,
+      "loss": 0.5459,
+      "step": 3954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940710576490516e-05,
+      "loss": 0.5691,
+      "step": 3955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406725547256437e-05,
+      "loss": 0.5201,
+      "step": 3956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406345211458713e-05,
+      "loss": 0.626,
+      "step": 3957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405964757516766e-05,
+      "loss": 0.5508,
+      "step": 3958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405584185435376e-05,
+      "loss": 0.5264,
+      "step": 3959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405203495219324e-05,
+      "loss": 0.5399,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404822686873387e-05,
+      "loss": 0.6047,
+      "step": 3961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940444176040235e-05,
+      "loss": 0.5871,
+      "step": 3962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404060715810996e-05,
+      "loss": 0.5935,
+      "step": 3963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403679553104117e-05,
+      "loss": 0.5779,
+      "step": 3964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403298272286492e-05,
+      "loss": 0.585,
+      "step": 3965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402916873362916e-05,
+      "loss": 0.5539,
+      "step": 3966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402535356338177e-05,
+      "loss": 0.6194,
+      "step": 3967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940215372121707e-05,
+      "loss": 0.5753,
+      "step": 3968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401771968004382e-05,
+      "loss": 0.6249,
+      "step": 3969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940139009670491e-05,
+      "loss": 0.6376,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401008107323457e-05,
+      "loss": 0.5674,
+      "step": 3971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400625999864813e-05,
+      "loss": 0.6094,
+      "step": 3972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940024377433378e-05,
+      "loss": 0.542,
+      "step": 3973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939986143073516e-05,
+      "loss": 0.6072,
+      "step": 3974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399478969073752e-05,
+      "loss": 0.5524,
+      "step": 3975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399096389354365e-05,
+      "loss": 0.5839,
+      "step": 3976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398713691581795e-05,
+      "loss": 0.6519,
+      "step": 3977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398330875760858e-05,
+      "loss": 0.5789,
+      "step": 3978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939794794189636e-05,
+      "loss": 0.5722,
+      "step": 3979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939756488999311e-05,
+      "loss": 0.5334,
+      "step": 3980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397181720055918e-05,
+      "loss": 0.5339,
+      "step": 3981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396798432089594e-05,
+      "loss": 0.6228,
+      "step": 3982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939641502609896e-05,
+      "loss": 0.6083,
+      "step": 3983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396031502088822e-05,
+      "loss": 0.592,
+      "step": 3984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395647860064007e-05,
+      "loss": 0.6054,
+      "step": 3985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395264100029325e-05,
+      "loss": 0.5705,
+      "step": 3986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394880221989604e-05,
+      "loss": 0.5963,
+      "step": 3987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939449622594966e-05,
+      "loss": 0.5615,
+      "step": 3988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394112111914314e-05,
+      "loss": 0.588,
+      "step": 3989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393727879888394e-05,
+      "loss": 0.5807,
+      "step": 3990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393343529876728e-05,
+      "loss": 0.6166,
+      "step": 3991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9392959061884142e-05,
+      "loss": 0.5618,
+      "step": 3992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9392574475915463e-05,
+      "loss": 0.6274,
+      "step": 3993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9392189771975522e-05,
+      "loss": 0.6041,
+      "step": 3994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939180495006915e-05,
+      "loss": 0.5986,
+      "step": 3995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391420010201184e-05,
+      "loss": 0.583,
+      "step": 3996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391034952376454e-05,
+      "loss": 0.6325,
+      "step": 3997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390649776599804e-05,
+      "loss": 0.5783,
+      "step": 3998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390264482876066e-05,
+      "loss": 0.5052,
+      "step": 3999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389879071210075e-05,
+      "loss": 0.6064,
+      "step": 4000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389493541606682e-05,
+      "loss": 0.582,
+      "step": 4001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389107894070723e-05,
+      "loss": 0.59,
+      "step": 4002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938872212860704e-05,
+      "loss": 0.5427,
+      "step": 4003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388336245220484e-05,
+      "loss": 0.524,
+      "step": 4004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93879502439159e-05,
+      "loss": 0.532,
+      "step": 4005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387564124698133e-05,
+      "loss": 0.554,
+      "step": 4006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387177887572036e-05,
+      "loss": 0.5218,
+      "step": 4007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938679153254246e-05,
+      "loss": 0.632,
+      "step": 4008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938640505961425e-05,
+      "loss": 0.5777,
+      "step": 4009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386018468792275e-05,
+      "loss": 0.5468,
+      "step": 4010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938563176008138e-05,
+      "loss": 0.5921,
+      "step": 4011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385244933486425e-05,
+      "loss": 0.6035,
+      "step": 4012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384857989012268e-05,
+      "loss": 0.5902,
+      "step": 4013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384470926663767e-05,
+      "loss": 0.6104,
+      "step": 4014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384083746445786e-05,
+      "loss": 0.5825,
+      "step": 4015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938369644836319e-05,
+      "loss": 0.6544,
+      "step": 4016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383309032420836e-05,
+      "loss": 0.6408,
+      "step": 4017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382921498623598e-05,
+      "loss": 0.6405,
+      "step": 4018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382533846976345e-05,
+      "loss": 0.5874,
+      "step": 4019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382146077483936e-05,
+      "loss": 0.5875,
+      "step": 4020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381758190151246e-05,
+      "loss": 0.5745,
+      "step": 4021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381370184983148e-05,
+      "loss": 0.5533,
+      "step": 4022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938098206198452e-05,
+      "loss": 0.6147,
+      "step": 4023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380593821160224e-05,
+      "loss": 0.6066,
+      "step": 4024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380205462515153e-05,
+      "loss": 0.609,
+      "step": 4025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937981698605417e-05,
+      "loss": 0.6626,
+      "step": 4026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379428391782163e-05,
+      "loss": 0.585,
+      "step": 4027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937903967970401e-05,
+      "loss": 0.5633,
+      "step": 4028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937865084982459e-05,
+      "loss": 0.6057,
+      "step": 4029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378261902148795e-05,
+      "loss": 0.5753,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377872836681503e-05,
+      "loss": 0.6551,
+      "step": 4031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93774836534276e-05,
+      "loss": 0.5888,
+      "step": 4032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937709435239198e-05,
+      "loss": 0.5547,
+      "step": 4033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376704933579533e-05,
+      "loss": 0.5637,
+      "step": 4034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937631539699514e-05,
+      "loss": 0.5258,
+      "step": 4035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375925742643703e-05,
+      "loss": 0.5504,
+      "step": 4036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375535970530117e-05,
+      "loss": 0.5608,
+      "step": 4037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375146080659274e-05,
+      "loss": 0.6008,
+      "step": 4038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937475607303607e-05,
+      "loss": 0.5526,
+      "step": 4039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93743659476654e-05,
+      "loss": 0.6039,
+      "step": 4040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373975704552173e-05,
+      "loss": 0.5906,
+      "step": 4041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373585343701283e-05,
+      "loss": 0.5402,
+      "step": 4042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937319486511764e-05,
+      "loss": 0.5827,
+      "step": 4043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372804268806146e-05,
+      "loss": 0.5867,
+      "step": 4044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93724135547717e-05,
+      "loss": 0.585,
+      "step": 4045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372022723019216e-05,
+      "loss": 0.5844,
+      "step": 4046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371631773553602e-05,
+      "loss": 0.5959,
+      "step": 4047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371240706379773e-05,
+      "loss": 0.5451,
+      "step": 4048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937084952150263e-05,
+      "loss": 0.6187,
+      "step": 4049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370458218927095e-05,
+      "loss": 0.5567,
+      "step": 4050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937006679865808e-05,
+      "loss": 0.5891,
+      "step": 4051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93696752607005e-05,
+      "loss": 0.5284,
+      "step": 4052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369283605059274e-05,
+      "loss": 0.5566,
+      "step": 4053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936889183173932e-05,
+      "loss": 0.5764,
+      "step": 4054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368499940745562e-05,
+      "loss": 0.6336,
+      "step": 4055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936810793208292e-05,
+      "loss": 0.5905,
+      "step": 4056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367715805756315e-05,
+      "loss": 0.5743,
+      "step": 4057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367323561770678e-05,
+      "loss": 0.5935,
+      "step": 4058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936693120013093e-05,
+      "loss": 0.57,
+      "step": 4059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366538720842005e-05,
+      "loss": 0.6165,
+      "step": 4060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366146123908828e-05,
+      "loss": 0.6105,
+      "step": 4061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936575340933633e-05,
+      "loss": 0.5925,
+      "step": 4062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365360577129442e-05,
+      "loss": 0.5945,
+      "step": 4063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364967627293107e-05,
+      "loss": 0.5728,
+      "step": 4064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936457455983225e-05,
+      "loss": 0.6251,
+      "step": 4065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364181374751812e-05,
+      "loss": 0.5396,
+      "step": 4066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936378807205673e-05,
+      "loss": 0.5644,
+      "step": 4067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936339465175195e-05,
+      "loss": 0.5906,
+      "step": 4068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363001113842404e-05,
+      "loss": 0.5696,
+      "step": 4069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362607458333044e-05,
+      "loss": 0.5974,
+      "step": 4070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362213685228805e-05,
+      "loss": 0.6264,
+      "step": 4071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936181979453464e-05,
+      "loss": 0.5424,
+      "step": 4072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361425786255496e-05,
+      "loss": 0.6177,
+      "step": 4073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936103166039632e-05,
+      "loss": 0.5641,
+      "step": 4074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360637416962058e-05,
+      "loss": 0.6091,
+      "step": 4075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936024305595767e-05,
+      "loss": 0.5983,
+      "step": 4076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93598485773881e-05,
+      "loss": 0.6076,
+      "step": 4077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359453981258313e-05,
+      "loss": 0.5772,
+      "step": 4078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935905926757326e-05,
+      "loss": 0.5469,
+      "step": 4079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358664436337896e-05,
+      "loss": 0.5675,
+      "step": 4080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935826948755718e-05,
+      "loss": 0.5352,
+      "step": 4081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935787442123608e-05,
+      "loss": 0.5592,
+      "step": 4082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357479237379552e-05,
+      "loss": 0.5295,
+      "step": 4083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357083935992557e-05,
+      "loss": 0.607,
+      "step": 4084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356688517080066e-05,
+      "loss": 0.4788,
+      "step": 4085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356292980647046e-05,
+      "loss": 0.6053,
+      "step": 4086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355897326698457e-05,
+      "loss": 0.5321,
+      "step": 4087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935550155523928e-05,
+      "loss": 0.6318,
+      "step": 4088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355105666274474e-05,
+      "loss": 0.5948,
+      "step": 4089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354709659809017e-05,
+      "loss": 0.5298,
+      "step": 4090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354313535847887e-05,
+      "loss": 0.5854,
+      "step": 4091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353917294396053e-05,
+      "loss": 0.6053,
+      "step": 4092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353520935458493e-05,
+      "loss": 0.6475,
+      "step": 4093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353124459040183e-05,
+      "loss": 0.5815,
+      "step": 4094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352727865146112e-05,
+      "loss": 0.5935,
+      "step": 4095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352331153781253e-05,
+      "loss": 0.5701,
+      "step": 4096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935193432495059e-05,
+      "loss": 0.6033,
+      "step": 4097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935153737865911e-05,
+      "loss": 0.6229,
+      "step": 4098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351140314911798e-05,
+      "loss": 0.6096,
+      "step": 4099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350743133713633e-05,
+      "loss": 0.57,
+      "step": 4100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350345835069616e-05,
+      "loss": 0.5752,
+      "step": 4101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349948418984733e-05,
+      "loss": 0.5579,
+      "step": 4102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934955088546397e-05,
+      "loss": 0.5599,
+      "step": 4103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934915323451233e-05,
+      "loss": 0.6063,
+      "step": 4104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348755466134794e-05,
+      "loss": 0.6762,
+      "step": 4105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934835758033637e-05,
+      "loss": 0.583,
+      "step": 4106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934795957712205e-05,
+      "loss": 0.5915,
+      "step": 4107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347561456496837e-05,
+      "loss": 0.5764,
+      "step": 4108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347163218465727e-05,
+      "loss": 0.5776,
+      "step": 4109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346764863033722e-05,
+      "loss": 0.6084,
+      "step": 4110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346366390205828e-05,
+      "loss": 0.5301,
+      "step": 4111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345967799987047e-05,
+      "loss": 0.5591,
+      "step": 4112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934556909238239e-05,
+      "loss": 0.537,
+      "step": 4113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345170267396863e-05,
+      "loss": 0.5354,
+      "step": 4114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934477132503547e-05,
+      "loss": 0.5883,
+      "step": 4115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344372265303225e-05,
+      "loss": 0.5708,
+      "step": 4116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343973088205145e-05,
+      "loss": 0.6328,
+      "step": 4117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934357379374624e-05,
+      "loss": 0.5941,
+      "step": 4118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934317438193152e-05,
+      "loss": 0.5952,
+      "step": 4119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934277485276601e-05,
+      "loss": 0.5422,
+      "step": 4120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342375206254723e-05,
+      "loss": 0.5419,
+      "step": 4121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341975442402683e-05,
+      "loss": 0.5743,
+      "step": 4122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341575561214906e-05,
+      "loss": 0.5722,
+      "step": 4123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341175562696417e-05,
+      "loss": 0.5916,
+      "step": 4124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934077544685224e-05,
+      "loss": 0.6656,
+      "step": 4125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93403752136874e-05,
+      "loss": 0.5888,
+      "step": 4126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339974863206926e-05,
+      "loss": 0.5462,
+      "step": 4127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933957439541584e-05,
+      "loss": 0.5373,
+      "step": 4128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339173810319177e-05,
+      "loss": 0.574,
+      "step": 4129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338773107921974e-05,
+      "loss": 0.5695,
+      "step": 4130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338372288229253e-05,
+      "loss": 0.5812,
+      "step": 4131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337971351246055e-05,
+      "loss": 0.5497,
+      "step": 4132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337570296977412e-05,
+      "loss": 0.6634,
+      "step": 4133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337169125428365e-05,
+      "loss": 0.607,
+      "step": 4134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933676783660395e-05,
+      "loss": 0.5706,
+      "step": 4135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933636643050921e-05,
+      "loss": 0.6388,
+      "step": 4136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933596490714918e-05,
+      "loss": 0.5815,
+      "step": 4137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933556326652891e-05,
+      "loss": 0.6269,
+      "step": 4138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335161508653443e-05,
+      "loss": 0.6164,
+      "step": 4139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334759633527828e-05,
+      "loss": 0.5829,
+      "step": 4140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93343576411571e-05,
+      "loss": 0.5817,
+      "step": 4141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333955531546326e-05,
+      "loss": 0.5658,
+      "step": 4142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933355330470055e-05,
+      "loss": 0.6279,
+      "step": 4143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333150960624812e-05,
+      "loss": 0.5825,
+      "step": 4144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933274849932418e-05,
+      "loss": 0.5393,
+      "step": 4145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332345920803704e-05,
+      "loss": 0.5444,
+      "step": 4146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331943225068443e-05,
+      "loss": 0.5386,
+      "step": 4147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933154041212345e-05,
+      "loss": 0.6095,
+      "step": 4148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331137481973785e-05,
+      "loss": 0.6089,
+      "step": 4149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330734434624512e-05,
+      "loss": 0.5923,
+      "step": 4150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330331270080692e-05,
+      "loss": 0.6511,
+      "step": 4151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932992798834739e-05,
+      "loss": 0.5929,
+      "step": 4152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932952458942967e-05,
+      "loss": 0.6439,
+      "step": 4153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329121073332597e-05,
+      "loss": 0.5516,
+      "step": 4154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328717440061243e-05,
+      "loss": 0.5413,
+      "step": 4155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932831368962067e-05,
+      "loss": 0.6396,
+      "step": 4156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932790982201596e-05,
+      "loss": 0.6113,
+      "step": 4157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932750583725218e-05,
+      "loss": 0.5883,
+      "step": 4158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327101735334404e-05,
+      "loss": 0.5877,
+      "step": 4159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326697516267704e-05,
+      "loss": 0.636,
+      "step": 4160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326293180057164e-05,
+      "loss": 0.5518,
+      "step": 4161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932588872670786e-05,
+      "loss": 0.6525,
+      "step": 4162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325484156224872e-05,
+      "loss": 0.5958,
+      "step": 4163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932507946861328e-05,
+      "loss": 0.595,
+      "step": 4164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324674663878163e-05,
+      "loss": 0.5725,
+      "step": 4165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324269742024615e-05,
+      "loss": 0.588,
+      "step": 4166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323864703057717e-05,
+      "loss": 0.5826,
+      "step": 4167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323459546982554e-05,
+      "loss": 0.5292,
+      "step": 4168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932305427380422e-05,
+      "loss": 0.5813,
+      "step": 4169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93226488835278e-05,
+      "loss": 0.6037,
+      "step": 4170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322243376158388e-05,
+      "loss": 0.5349,
+      "step": 4171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321837751701076e-05,
+      "loss": 0.6407,
+      "step": 4172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932143201016096e-05,
+      "loss": 0.539,
+      "step": 4173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321026151543133e-05,
+      "loss": 0.5673,
+      "step": 4174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93206201758527e-05,
+      "loss": 0.5946,
+      "step": 4175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320214083094756e-05,
+      "loss": 0.6121,
+      "step": 4176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93198078732744e-05,
+      "loss": 0.5477,
+      "step": 4177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319401546396735e-05,
+      "loss": 0.5238,
+      "step": 4178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318995102466865e-05,
+      "loss": 0.5791,
+      "step": 4179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318588541489893e-05,
+      "loss": 0.5838,
+      "step": 4180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318181863470928e-05,
+      "loss": 0.607,
+      "step": 4181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317775068415076e-05,
+      "loss": 0.6072,
+      "step": 4182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931736815632745e-05,
+      "loss": 0.6033,
+      "step": 4183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316961127213156e-05,
+      "loss": 0.601,
+      "step": 4184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931655398107731e-05,
+      "loss": 0.5767,
+      "step": 4185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316146717925023e-05,
+      "loss": 0.649,
+      "step": 4186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931573933776141e-05,
+      "loss": 0.5994,
+      "step": 4187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931533184059159e-05,
+      "loss": 0.5648,
+      "step": 4188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931492422642068e-05,
+      "loss": 0.6702,
+      "step": 4189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9314516495253805e-05,
+      "loss": 0.5686,
+      "step": 4190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314108647096076e-05,
+      "loss": 0.6036,
+      "step": 4191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931370068195262e-05,
+      "loss": 0.5955,
+      "step": 4192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313292599828565e-05,
+      "loss": 0.6256,
+      "step": 4193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931288440072903e-05,
+      "loss": 0.5302,
+      "step": 4194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931247608465915e-05,
+      "loss": 0.6304,
+      "step": 4195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312067651624046e-05,
+      "loss": 0.5727,
+      "step": 4196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931165910162885e-05,
+      "loss": 0.6099,
+      "step": 4197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311250434678697e-05,
+      "loss": 0.5915,
+      "step": 4198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310841650778714e-05,
+      "loss": 0.5911,
+      "step": 4199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310432749934042e-05,
+      "loss": 0.5666,
+      "step": 4200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310023732149807e-05,
+      "loss": 0.6381,
+      "step": 4201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309614597431156e-05,
+      "loss": 0.5519,
+      "step": 4202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309205345783225e-05,
+      "loss": 0.5573,
+      "step": 4203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308795977211154e-05,
+      "loss": 0.5702,
+      "step": 4204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308386491720083e-05,
+      "loss": 0.5955,
+      "step": 4205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307976889315158e-05,
+      "loss": 0.5492,
+      "step": 4206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930756717000152e-05,
+      "loss": 0.6033,
+      "step": 4207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307157333784317e-05,
+      "loss": 0.561,
+      "step": 4208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306747380668697e-05,
+      "loss": 0.5954,
+      "step": 4209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930633731065981e-05,
+      "loss": 0.5419,
+      "step": 4210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93059271237628e-05,
+      "loss": 0.5935,
+      "step": 4211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305516819982832e-05,
+      "loss": 0.5657,
+      "step": 4212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930510639932505e-05,
+      "loss": 0.6062,
+      "step": 4213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304695861794605e-05,
+      "loss": 0.5787,
+      "step": 4214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304285207396664e-05,
+      "loss": 0.6244,
+      "step": 4215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303874436136376e-05,
+      "loss": 0.6222,
+      "step": 4216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303463548018907e-05,
+      "loss": 0.5831,
+      "step": 4217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303052543049414e-05,
+      "loss": 0.6109,
+      "step": 4218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930264142123306e-05,
+      "loss": 0.6649,
+      "step": 4219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930223018257501e-05,
+      "loss": 0.5952,
+      "step": 4220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301818827080423e-05,
+      "loss": 0.6364,
+      "step": 4221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301407354754475e-05,
+      "loss": 0.523,
+      "step": 4222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300995765602327e-05,
+      "loss": 0.6277,
+      "step": 4223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300584059629153e-05,
+      "loss": 0.5622,
+      "step": 4224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930017223684012e-05,
+      "loss": 0.6178,
+      "step": 4225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299760297240407e-05,
+      "loss": 0.5795,
+      "step": 4226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299348240835182e-05,
+      "loss": 0.661,
+      "step": 4227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298936067629622e-05,
+      "loss": 0.5605,
+      "step": 4228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298523777628905e-05,
+      "loss": 0.5538,
+      "step": 4229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298111370838208e-05,
+      "loss": 0.5375,
+      "step": 4230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929769884726271e-05,
+      "loss": 0.6062,
+      "step": 4231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92972862069076e-05,
+      "loss": 0.599,
+      "step": 4232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296873449778047e-05,
+      "loss": 0.5304,
+      "step": 4233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296460575879244e-05,
+      "loss": 0.5903,
+      "step": 4234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296047585216376e-05,
+      "loss": 0.6137,
+      "step": 4235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929563447779463e-05,
+      "loss": 0.5531,
+      "step": 4236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295221253619197e-05,
+      "loss": 0.6234,
+      "step": 4237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294807912695263e-05,
+      "loss": 0.597,
+      "step": 4238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294394455028017e-05,
+      "loss": 0.6104,
+      "step": 4239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929398088062266e-05,
+      "loss": 0.5154,
+      "step": 4240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293567189484385e-05,
+      "loss": 0.6456,
+      "step": 4241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929315338161838e-05,
+      "loss": 0.5661,
+      "step": 4242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292739457029853e-05,
+      "loss": 0.5274,
+      "step": 4243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292325415723997e-05,
+      "loss": 0.6037,
+      "step": 4244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929191125770601e-05,
+      "loss": 0.5746,
+      "step": 4245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92914969829811e-05,
+      "loss": 0.6513,
+      "step": 4246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291082591554468e-05,
+      "loss": 0.5631,
+      "step": 4247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290668083431313e-05,
+      "loss": 0.5313,
+      "step": 4248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290253458616854e-05,
+      "loss": 0.6329,
+      "step": 4249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289838717116288e-05,
+      "loss": 0.6336,
+      "step": 4250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928942385893483e-05,
+      "loss": 0.5572,
+      "step": 4251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289008884077682e-05,
+      "loss": 0.5623,
+      "step": 4252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288593792550065e-05,
+      "loss": 0.5463,
+      "step": 4253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928817858435719e-05,
+      "loss": 0.487,
+      "step": 4254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287763259504272e-05,
+      "loss": 0.633,
+      "step": 4255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928734781799653e-05,
+      "loss": 0.5948,
+      "step": 4256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286932259839175e-05,
+      "loss": 0.6342,
+      "step": 4257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928651658503743e-05,
+      "loss": 0.5841,
+      "step": 4258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286100793596523e-05,
+      "loss": 0.6345,
+      "step": 4259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285684885521664e-05,
+      "loss": 0.5385,
+      "step": 4260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285268860818086e-05,
+      "loss": 0.5562,
+      "step": 4261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928485271949101e-05,
+      "loss": 0.5921,
+      "step": 4262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928443646154566e-05,
+      "loss": 0.5378,
+      "step": 4263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284020086987273e-05,
+      "loss": 0.5733,
+      "step": 4264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283603595821076e-05,
+      "loss": 0.5249,
+      "step": 4265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928318698805229e-05,
+      "loss": 0.5899,
+      "step": 4266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928277026368616e-05,
+      "loss": 0.6437,
+      "step": 4267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282353422727915e-05,
+      "loss": 0.589,
+      "step": 4268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281936465182794e-05,
+      "loss": 0.5656,
+      "step": 4269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928151939105603e-05,
+      "loss": 0.5792,
+      "step": 4270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928110220035286e-05,
+      "loss": 0.6282,
+      "step": 4271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928068489307853e-05,
+      "loss": 0.6192,
+      "step": 4272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280267469238274e-05,
+      "loss": 0.575,
+      "step": 4273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279849928837338e-05,
+      "loss": 0.59,
+      "step": 4274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279432271880972e-05,
+      "loss": 0.6018,
+      "step": 4275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279014498374414e-05,
+      "loss": 0.6113,
+      "step": 4276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278596608322916e-05,
+      "loss": 0.5832,
+      "step": 4277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278178601731722e-05,
+      "loss": 0.4978,
+      "step": 4278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277760478606088e-05,
+      "loss": 0.5959,
+      "step": 4279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927734223895126e-05,
+      "loss": 0.5342,
+      "step": 4280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276923882772497e-05,
+      "loss": 0.6314,
+      "step": 4281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927650541007505e-05,
+      "loss": 0.5847,
+      "step": 4282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276086820864174e-05,
+      "loss": 0.5533,
+      "step": 4283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927566811514513e-05,
+      "loss": 0.5556,
+      "step": 4284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275249292923172e-05,
+      "loss": 0.641,
+      "step": 4285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9274830354203567e-05,
+      "loss": 0.6049,
+      "step": 4286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274411298991568e-05,
+      "loss": 0.5405,
+      "step": 4287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927399212729245e-05,
+      "loss": 0.5454,
+      "step": 4288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927357283911147e-05,
+      "loss": 0.5533,
+      "step": 4289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273153434453893e-05,
+      "loss": 0.5873,
+      "step": 4290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927273391332499e-05,
+      "loss": 0.5759,
+      "step": 4291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927231427573003e-05,
+      "loss": 0.5474,
+      "step": 4292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271894521674285e-05,
+      "loss": 0.5726,
+      "step": 4293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271474651163025e-05,
+      "loss": 0.619,
+      "step": 4294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271054664201523e-05,
+      "loss": 0.5412,
+      "step": 4295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270634560795054e-05,
+      "loss": 0.5546,
+      "step": 4296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270214340948897e-05,
+      "loss": 0.6404,
+      "step": 4297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926979400466833e-05,
+      "loss": 0.6264,
+      "step": 4298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269373551958634e-05,
+      "loss": 0.5219,
+      "step": 4299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926895298282508e-05,
+      "loss": 0.5806,
+      "step": 4300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268532297272963e-05,
+      "loss": 0.5997,
+      "step": 4301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268111495307557e-05,
+      "loss": 0.5684,
+      "step": 4302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267690576934152e-05,
+      "loss": 0.632,
+      "step": 4303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267269542158037e-05,
+      "loss": 0.553,
+      "step": 4304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266848390984497e-05,
+      "loss": 0.6084,
+      "step": 4305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266427123418822e-05,
+      "loss": 0.5654,
+      "step": 4306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266005739466304e-05,
+      "loss": 0.7101,
+      "step": 4307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265584239132234e-05,
+      "loss": 0.6039,
+      "step": 4308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265162622421906e-05,
+      "loss": 0.5912,
+      "step": 4309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264740889340618e-05,
+      "loss": 0.5512,
+      "step": 4310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264319039893663e-05,
+      "loss": 0.5735,
+      "step": 4311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263897074086345e-05,
+      "loss": 0.5809,
+      "step": 4312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926347499192396e-05,
+      "loss": 0.5185,
+      "step": 4313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926305279341181e-05,
+      "loss": 0.6031,
+      "step": 4314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262630478555197e-05,
+      "loss": 0.5248,
+      "step": 4315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262208047359427e-05,
+      "loss": 0.5827,
+      "step": 4316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261785499829807e-05,
+      "loss": 0.5936,
+      "step": 4317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261362835971643e-05,
+      "loss": 0.6273,
+      "step": 4318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260940055790242e-05,
+      "loss": 0.5947,
+      "step": 4319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260517159290915e-05,
+      "loss": 0.6061,
+      "step": 4320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260094146478972e-05,
+      "loss": 0.6386,
+      "step": 4321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925967101735973e-05,
+      "loss": 0.5703,
+      "step": 4322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92592477719385e-05,
+      "loss": 0.6167,
+      "step": 4323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92588244102206e-05,
+      "loss": 0.604,
+      "step": 4324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258400932211348e-05,
+      "loss": 0.5954,
+      "step": 4325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925797733791606e-05,
+      "loss": 0.559,
+      "step": 4326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925755362734006e-05,
+      "loss": 0.5555,
+      "step": 4327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257129800488665e-05,
+      "loss": 0.5873,
+      "step": 4328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256705857367205e-05,
+      "loss": 0.5713,
+      "step": 4329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256281797980997e-05,
+      "loss": 0.5486,
+      "step": 4330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255857622335373e-05,
+      "loss": 0.5595,
+      "step": 4331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925543333043566e-05,
+      "loss": 0.5873,
+      "step": 4332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255008922287186e-05,
+      "loss": 0.5893,
+      "step": 4333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254584397895282e-05,
+      "loss": 0.555,
+      "step": 4334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254159757265277e-05,
+      "loss": 0.5485,
+      "step": 4335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925373500040251e-05,
+      "loss": 0.6558,
+      "step": 4336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253310127312313e-05,
+      "loss": 0.5434,
+      "step": 4337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925288513800002e-05,
+      "loss": 0.5624,
+      "step": 4338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925246003247097e-05,
+      "loss": 0.6003,
+      "step": 4339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925203481073051e-05,
+      "loss": 0.595,
+      "step": 4340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925160947278397e-05,
+      "loss": 0.5808,
+      "step": 4341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92511840186367e-05,
+      "loss": 0.6182,
+      "step": 4342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250758448294034e-05,
+      "loss": 0.594,
+      "step": 4343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250332761761333e-05,
+      "loss": 0.5787,
+      "step": 4344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249906959043927e-05,
+      "loss": 0.645,
+      "step": 4345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249481040147177e-05,
+      "loss": 0.5595,
+      "step": 4346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249055005076426e-05,
+      "loss": 0.5708,
+      "step": 4347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924862885383702e-05,
+      "loss": 0.5067,
+      "step": 4348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248202586434325e-05,
+      "loss": 0.5643,
+      "step": 4349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247776202873684e-05,
+      "loss": 0.609,
+      "step": 4350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924734970316046e-05,
+      "loss": 0.5771,
+      "step": 4351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92469230873e-05,
+      "loss": 0.5276,
+      "step": 4352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246496355297672e-05,
+      "loss": 0.5515,
+      "step": 4353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924606950715883e-05,
+      "loss": 0.5727,
+      "step": 4354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245642542888836e-05,
+      "loss": 0.5469,
+      "step": 4355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245215462493055e-05,
+      "loss": 0.5929,
+      "step": 4356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244788265976853e-05,
+      "loss": 0.5664,
+      "step": 4357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924436095334559e-05,
+      "loss": 0.5395,
+      "step": 4358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243933524604636e-05,
+      "loss": 0.6276,
+      "step": 4359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924350597975936e-05,
+      "loss": 0.585,
+      "step": 4360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924307831881513e-05,
+      "loss": 0.5911,
+      "step": 4361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242650541777318e-05,
+      "loss": 0.6437,
+      "step": 4362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92422226486513e-05,
+      "loss": 0.5593,
+      "step": 4363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241794639442448e-05,
+      "loss": 0.6237,
+      "step": 4364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241366514156133e-05,
+      "loss": 0.5591,
+      "step": 4365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240938272797743e-05,
+      "loss": 0.6118,
+      "step": 4366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240509915372647e-05,
+      "loss": 0.5717,
+      "step": 4367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924008144188623e-05,
+      "loss": 0.6158,
+      "step": 4368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239652852343874e-05,
+      "loss": 0.5392,
+      "step": 4369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239224146750958e-05,
+      "loss": 0.6884,
+      "step": 4370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.5066,
+      "step": 4371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238366387434993e-05,
+      "loss": 0.563,
+      "step": 4372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237937333722723e-05,
+      "loss": 0.6194,
+      "step": 4373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237508163981437e-05,
+      "loss": 0.6439,
+      "step": 4374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237078878216535e-05,
+      "loss": 0.5667,
+      "step": 4375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236649476433402e-05,
+      "loss": 0.6044,
+      "step": 4376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236219958637435e-05,
+      "loss": 0.6214,
+      "step": 4377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923579032483403e-05,
+      "loss": 0.5653,
+      "step": 4378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923536057502858e-05,
+      "loss": 0.6083,
+      "step": 4379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234930709226484e-05,
+      "loss": 0.5875,
+      "step": 4380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234500727433142e-05,
+      "loss": 0.6086,
+      "step": 4381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923407062965395e-05,
+      "loss": 0.6445,
+      "step": 4382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9233640415894317e-05,
+      "loss": 0.7048,
+      "step": 4383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923321008615964e-05,
+      "loss": 0.5725,
+      "step": 4384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232779640455332e-05,
+      "loss": 0.5506,
+      "step": 4385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232349078786793e-05,
+      "loss": 0.6071,
+      "step": 4386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231918401159433e-05,
+      "loss": 0.5886,
+      "step": 4387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923148760757866e-05,
+      "loss": 0.574,
+      "step": 4388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231056698049883e-05,
+      "loss": 0.634,
+      "step": 4389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923062567257852e-05,
+      "loss": 0.6167,
+      "step": 4390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230194531169978e-05,
+      "loss": 0.5908,
+      "step": 4391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922976327382968e-05,
+      "loss": 0.5923,
+      "step": 4392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229331900563038e-05,
+      "loss": 0.5386,
+      "step": 4393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228900411375467e-05,
+      "loss": 0.564,
+      "step": 4394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228468806272393e-05,
+      "loss": 0.5519,
+      "step": 4395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228037085259234e-05,
+      "loss": 0.5482,
+      "step": 4396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227605248341413e-05,
+      "loss": 0.577,
+      "step": 4397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922717329552435e-05,
+      "loss": 0.6009,
+      "step": 4398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226741226813477e-05,
+      "loss": 0.5268,
+      "step": 4399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226309042214215e-05,
+      "loss": 0.6534,
+      "step": 4400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225876741731996e-05,
+      "loss": 0.542,
+      "step": 4401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225444325372248e-05,
+      "loss": 0.5697,
+      "step": 4402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92250117931404e-05,
+      "loss": 0.6692,
+      "step": 4403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922457914504189e-05,
+      "loss": 0.5569,
+      "step": 4404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224146381082153e-05,
+      "loss": 0.6046,
+      "step": 4405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223713501266616e-05,
+      "loss": 0.5939,
+      "step": 4406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922328050560072e-05,
+      "loss": 0.5624,
+      "step": 4407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222847394089906e-05,
+      "loss": 0.5705,
+      "step": 4408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222414166739614e-05,
+      "loss": 0.6214,
+      "step": 4409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221980823555283e-05,
+      "loss": 0.5821,
+      "step": 4410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221547364542354e-05,
+      "loss": 0.5337,
+      "step": 4411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221113789706273e-05,
+      "loss": 0.5238,
+      "step": 4412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922068009905249e-05,
+      "loss": 0.5647,
+      "step": 4413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220246292586446e-05,
+      "loss": 0.5716,
+      "step": 4414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219812370313594e-05,
+      "loss": 0.5458,
+      "step": 4415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921937833223938e-05,
+      "loss": 0.643,
+      "step": 4416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921894417836926e-05,
+      "loss": 0.5152,
+      "step": 4417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218509908708683e-05,
+      "loss": 0.551,
+      "step": 4418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218075523263104e-05,
+      "loss": 0.617,
+      "step": 4419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217641022037984e-05,
+      "loss": 0.6131,
+      "step": 4420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217206405038774e-05,
+      "loss": 0.6034,
+      "step": 4421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216771672270937e-05,
+      "loss": 0.6473,
+      "step": 4422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921633682373993e-05,
+      "loss": 0.5367,
+      "step": 4423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921590185945122e-05,
+      "loss": 0.5206,
+      "step": 4424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921546677941026e-05,
+      "loss": 0.5747,
+      "step": 4425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215031583622526e-05,
+      "loss": 0.5488,
+      "step": 4426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214596272093478e-05,
+      "loss": 0.6152,
+      "step": 4427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214160844828585e-05,
+      "loss": 0.5647,
+      "step": 4428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213725301833313e-05,
+      "loss": 0.5906,
+      "step": 4429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921328964311314e-05,
+      "loss": 0.535,
+      "step": 4430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921285386867353e-05,
+      "loss": 0.6003,
+      "step": 4431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212417978519958e-05,
+      "loss": 0.5898,
+      "step": 4432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211981972657904e-05,
+      "loss": 0.5425,
+      "step": 4433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211545851092838e-05,
+      "loss": 0.6146,
+      "step": 4434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921110961383024e-05,
+      "loss": 0.6076,
+      "step": 4435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210673260875592e-05,
+      "loss": 0.6173,
+      "step": 4436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921023679223437e-05,
+      "loss": 0.5629,
+      "step": 4437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920980020791206e-05,
+      "loss": 0.577,
+      "step": 4438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920936350791414e-05,
+      "loss": 0.5749,
+      "step": 4439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92089266922461e-05,
+      "loss": 0.5579,
+      "step": 4440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208489760913424e-05,
+      "loss": 0.6505,
+      "step": 4441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208052713921606e-05,
+      "loss": 0.5302,
+      "step": 4442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207615551276123e-05,
+      "loss": 0.6527,
+      "step": 4443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207178272982477e-05,
+      "loss": 0.6354,
+      "step": 4444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206740879046157e-05,
+      "loss": 0.5956,
+      "step": 4445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206303369472655e-05,
+      "loss": 0.5621,
+      "step": 4446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205865744267467e-05,
+      "loss": 0.5859,
+      "step": 4447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205428003436088e-05,
+      "loss": 0.6077,
+      "step": 4448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920499014698402e-05,
+      "loss": 0.5221,
+      "step": 4449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920455217491676e-05,
+      "loss": 0.5674,
+      "step": 4450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204114087239806e-05,
+      "loss": 0.6087,
+      "step": 4451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203675883958666e-05,
+      "loss": 0.5992,
+      "step": 4452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203237565078844e-05,
+      "loss": 0.5999,
+      "step": 4453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920279913060584e-05,
+      "loss": 0.626,
+      "step": 4454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202360580545166e-05,
+      "loss": 0.5322,
+      "step": 4455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201921914902323e-05,
+      "loss": 0.5592,
+      "step": 4456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920148313368283e-05,
+      "loss": 0.5919,
+      "step": 4457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920104423689219e-05,
+      "loss": 0.6744,
+      "step": 4458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920060522453592e-05,
+      "loss": 0.5667,
+      "step": 4459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200166096619534e-05,
+      "loss": 0.5642,
+      "step": 4460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199726853148545e-05,
+      "loss": 0.6155,
+      "step": 4461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199287494128473e-05,
+      "loss": 0.6449,
+      "step": 4462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198848019564833e-05,
+      "loss": 0.6068,
+      "step": 4463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198408429463147e-05,
+      "loss": 0.5686,
+      "step": 4464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919796872382894e-05,
+      "loss": 0.6072,
+      "step": 4465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919752890266773e-05,
+      "loss": 0.6676,
+      "step": 4466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197088965985037e-05,
+      "loss": 0.5595,
+      "step": 4467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196648913786394e-05,
+      "loss": 0.5492,
+      "step": 4468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196208746077326e-05,
+      "loss": 0.5823,
+      "step": 4469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919576846286336e-05,
+      "loss": 0.605,
+      "step": 4470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919532806415003e-05,
+      "loss": 0.6083,
+      "step": 4471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194887549942862e-05,
+      "loss": 0.6258,
+      "step": 4472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919444692024739e-05,
+      "loss": 0.5994,
+      "step": 4473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194006175069153e-05,
+      "loss": 0.5924,
+      "step": 4474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919356531441368e-05,
+      "loss": 0.5658,
+      "step": 4475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193124338286514e-05,
+      "loss": 0.6234,
+      "step": 4476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192683246693193e-05,
+      "loss": 0.5427,
+      "step": 4477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192242039639252e-05,
+      "loss": 0.613,
+      "step": 4478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919180071713024e-05,
+      "loss": 0.5182,
+      "step": 4479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191359279171694e-05,
+      "loss": 0.5084,
+      "step": 4480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190917725769162e-05,
+      "loss": 0.6102,
+      "step": 4481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190476056928186e-05,
+      "loss": 0.5226,
+      "step": 4482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190034272654316e-05,
+      "loss": 0.5888,
+      "step": 4483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189592372953104e-05,
+      "loss": 0.6091,
+      "step": 4484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189150357830094e-05,
+      "loss": 0.5682,
+      "step": 4485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188708227290842e-05,
+      "loss": 0.5172,
+      "step": 4486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188265981340902e-05,
+      "loss": 0.6045,
+      "step": 4487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187823619985826e-05,
+      "loss": 0.4947,
+      "step": 4488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187381143231167e-05,
+      "loss": 0.6106,
+      "step": 4489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918693855108249e-05,
+      "loss": 0.5835,
+      "step": 4490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918649584354535e-05,
+      "loss": 0.5493,
+      "step": 4491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186053020625307e-05,
+      "loss": 0.6158,
+      "step": 4492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185610082327925e-05,
+      "loss": 0.6088,
+      "step": 4493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185167028658764e-05,
+      "loss": 0.5591,
+      "step": 4494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918472385962339e-05,
+      "loss": 0.6018,
+      "step": 4495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184280575227373e-05,
+      "loss": 0.6215,
+      "step": 4496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183837175476274e-05,
+      "loss": 0.5691,
+      "step": 4497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183393660375665e-05,
+      "loss": 0.568,
+      "step": 4498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918295002993112e-05,
+      "loss": 0.5961,
+      "step": 4499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918250628414821e-05,
+      "loss": 0.5661,
+      "step": 4500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182062423032506e-05,
+      "loss": 0.567,
+      "step": 4501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181618446589582e-05,
+      "loss": 0.6432,
+      "step": 4502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918117435482502e-05,
+      "loss": 0.5686,
+      "step": 4503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918073014774439e-05,
+      "loss": 0.6119,
+      "step": 4504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180285825353273e-05,
+      "loss": 0.58,
+      "step": 4505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179841387657254e-05,
+      "loss": 0.5862,
+      "step": 4506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179396834661913e-05,
+      "loss": 0.6279,
+      "step": 4507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178952166372835e-05,
+      "loss": 0.6021,
+      "step": 4508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178507382795604e-05,
+      "loss": 0.55,
+      "step": 4509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178062483935806e-05,
+      "loss": 0.5872,
+      "step": 4510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917761746979903e-05,
+      "loss": 0.6044,
+      "step": 4511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177172340390863e-05,
+      "loss": 0.6187,
+      "step": 4512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176727095716898e-05,
+      "loss": 0.5584,
+      "step": 4513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176281735782725e-05,
+      "loss": 0.5997,
+      "step": 4514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175836260593937e-05,
+      "loss": 0.6256,
+      "step": 4515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917539067015614e-05,
+      "loss": 0.5801,
+      "step": 4516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174944964474914e-05,
+      "loss": 0.6273,
+      "step": 4517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917449914355587e-05,
+      "loss": 0.6009,
+      "step": 4518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91740532074046e-05,
+      "loss": 0.6022,
+      "step": 4519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917360715602671e-05,
+      "loss": 0.5655,
+      "step": 4520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91731609894278e-05,
+      "loss": 0.5626,
+      "step": 4521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172714707613474e-05,
+      "loss": 0.6259,
+      "step": 4522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917226831058934e-05,
+      "loss": 0.5778,
+      "step": 4523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171821798361e-05,
+      "loss": 0.5999,
+      "step": 4524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171375170934063e-05,
+      "loss": 0.5927,
+      "step": 4525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170928428314142e-05,
+      "loss": 0.583,
+      "step": 4526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170481570506848e-05,
+      "loss": 0.613,
+      "step": 4527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917003459751779e-05,
+      "loss": 0.662,
+      "step": 4528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916958750935258e-05,
+      "loss": 0.5442,
+      "step": 4529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169140306016848e-05,
+      "loss": 0.5314,
+      "step": 4530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916869298751619e-05,
+      "loss": 0.508,
+      "step": 4531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916824555385624e-05,
+      "loss": 0.577,
+      "step": 4532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167798005042608e-05,
+      "loss": 0.5352,
+      "step": 4533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167350341080922e-05,
+      "loss": 0.4507,
+      "step": 4534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166902561976806e-05,
+      "loss": 0.5054,
+      "step": 4535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166454667735874e-05,
+      "loss": 0.6351,
+      "step": 4536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166006658363764e-05,
+      "loss": 0.5498,
+      "step": 4537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165558533866093e-05,
+      "loss": 0.5444,
+      "step": 4538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165110294248493e-05,
+      "loss": 0.5112,
+      "step": 4539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164661939516595e-05,
+      "loss": 0.5667,
+      "step": 4540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916421346967603e-05,
+      "loss": 0.5244,
+      "step": 4541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916376488473243e-05,
+      "loss": 0.6,
+      "step": 4542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916331618469143e-05,
+      "loss": 0.6044,
+      "step": 4543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916286736955867e-05,
+      "loss": 0.5619,
+      "step": 4544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162418439339778e-05,
+      "loss": 0.5977,
+      "step": 4545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161969394040397e-05,
+      "loss": 0.5499,
+      "step": 4546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161520233666172e-05,
+      "loss": 0.6029,
+      "step": 4547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161070958222735e-05,
+      "loss": 0.5665,
+      "step": 4548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160621567715733e-05,
+      "loss": 0.5137,
+      "step": 4549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160172062150816e-05,
+      "loss": 0.607,
+      "step": 4550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159722441533622e-05,
+      "loss": 0.5507,
+      "step": 4551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91592727058698e-05,
+      "loss": 0.5846,
+      "step": 4552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158822855165003e-05,
+      "loss": 0.5403,
+      "step": 4553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158372889424872e-05,
+      "loss": 0.5436,
+      "step": 4554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157922808655072e-05,
+      "loss": 0.5437,
+      "step": 4555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157472612861244e-05,
+      "loss": 0.649,
+      "step": 4556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157022302049045e-05,
+      "loss": 0.5903,
+      "step": 4557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156571876224138e-05,
+      "loss": 0.5486,
+      "step": 4558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156121335392174e-05,
+      "loss": 0.5166,
+      "step": 4559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155670679558814e-05,
+      "loss": 0.5492,
+      "step": 4560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155219908729717e-05,
+      "loss": 0.5741,
+      "step": 4561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154769022910544e-05,
+      "loss": 0.5522,
+      "step": 4562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915431802210696e-05,
+      "loss": 0.6386,
+      "step": 4563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153866906324627e-05,
+      "loss": 0.6483,
+      "step": 4564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915341567556922e-05,
+      "loss": 0.5371,
+      "step": 4565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915296432984639e-05,
+      "loss": 0.5026,
+      "step": 4566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915251286916182e-05,
+      "loss": 0.6098,
+      "step": 4567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152061293521174e-05,
+      "loss": 0.5778,
+      "step": 4568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915160960293013e-05,
+      "loss": 0.5984,
+      "step": 4569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151157797394353e-05,
+      "loss": 0.6254,
+      "step": 4570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150705876919522e-05,
+      "loss": 0.5905,
+      "step": 4571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150253841511316e-05,
+      "loss": 0.5128,
+      "step": 4572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149801691175406e-05,
+      "loss": 0.6303,
+      "step": 4573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914934942591747e-05,
+      "loss": 0.5289,
+      "step": 4574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91488970457432e-05,
+      "loss": 0.6525,
+      "step": 4575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914844455065827e-05,
+      "loss": 0.5781,
+      "step": 4576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914799194066836e-05,
+      "loss": 0.5926,
+      "step": 4577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914753921577916e-05,
+      "loss": 0.5414,
+      "step": 4578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914708637599636e-05,
+      "loss": 0.602,
+      "step": 4579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146633421325637e-05,
+      "loss": 0.5439,
+      "step": 4580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146180351772686e-05,
+      "loss": 0.6029,
+      "step": 4581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145727167343203e-05,
+      "loss": 0.6226,
+      "step": 4582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914527386804287e-05,
+      "loss": 0.6086,
+      "step": 4583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144820453877385e-05,
+      "loss": 0.5687,
+      "step": 4584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144366924852445e-05,
+      "loss": 0.5846,
+      "step": 4585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143913280973745e-05,
+      "loss": 0.5905,
+      "step": 4586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143459522246978e-05,
+      "loss": 0.5346,
+      "step": 4587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914300564867785e-05,
+      "loss": 0.5973,
+      "step": 4588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142551660272063e-05,
+      "loss": 0.5698,
+      "step": 4589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914209755703531e-05,
+      "loss": 0.5217,
+      "step": 4590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91416433389733e-05,
+      "loss": 0.5804,
+      "step": 4591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9141189006091738e-05,
+      "loss": 0.6544,
+      "step": 4592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914073455839633e-05,
+      "loss": 0.6031,
+      "step": 4593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140279995892787e-05,
+      "loss": 0.6282,
+      "step": 4594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139825318586812e-05,
+      "loss": 0.61,
+      "step": 4595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913937052648412e-05,
+      "loss": 0.6929,
+      "step": 4596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913891561959042e-05,
+      "loss": 0.5951,
+      "step": 4597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138460597911427e-05,
+      "loss": 0.5604,
+      "step": 4598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138005461452863e-05,
+      "loss": 0.4998,
+      "step": 4599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137550210220436e-05,
+      "loss": 0.5768,
+      "step": 4600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137094844219867e-05,
+      "loss": 0.6161,
+      "step": 4601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136639363456873e-05,
+      "loss": 0.6168,
+      "step": 4602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913618376793718e-05,
+      "loss": 0.6311,
+      "step": 4603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135728057666505e-05,
+      "loss": 0.5419,
+      "step": 4604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135272232650575e-05,
+      "loss": 0.6252,
+      "step": 4605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134816292895112e-05,
+      "loss": 0.5633,
+      "step": 4606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134360238405847e-05,
+      "loss": 0.6409,
+      "step": 4607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133904069188505e-05,
+      "loss": 0.5391,
+      "step": 4608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133447785248816e-05,
+      "loss": 0.6504,
+      "step": 4609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132991386592512e-05,
+      "loss": 0.598,
+      "step": 4610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132534873225323e-05,
+      "loss": 0.5644,
+      "step": 4611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132078245152988e-05,
+      "loss": 0.5808,
+      "step": 4612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131621502381237e-05,
+      "loss": 0.609,
+      "step": 4613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913116464491581e-05,
+      "loss": 0.4616,
+      "step": 4614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913070767276244e-05,
+      "loss": 0.5426,
+      "step": 4615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913025058592687e-05,
+      "loss": 0.5826,
+      "step": 4616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129793384414846e-05,
+      "loss": 0.607,
+      "step": 4617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129336068232104e-05,
+      "loss": 0.5624,
+      "step": 4618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128878637384387e-05,
+      "loss": 0.6195,
+      "step": 4619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128421091877446e-05,
+      "loss": 0.5614,
+      "step": 4620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127963431717026e-05,
+      "loss": 0.5635,
+      "step": 4621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912750565690887e-05,
+      "loss": 0.6468,
+      "step": 4622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127047767458732e-05,
+      "loss": 0.5715,
+      "step": 4623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126589763372364e-05,
+      "loss": 0.5705,
+      "step": 4624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912613164465552e-05,
+      "loss": 0.5338,
+      "step": 4625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912567341131395e-05,
+      "loss": 0.583,
+      "step": 4626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125215063353408e-05,
+      "loss": 0.5481,
+      "step": 4627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124756600779656e-05,
+      "loss": 0.579,
+      "step": 4628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912429802359845e-05,
+      "loss": 0.5396,
+      "step": 4629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912383933181555e-05,
+      "loss": 0.557,
+      "step": 4630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912338052543672e-05,
+      "loss": 0.5619,
+      "step": 4631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122921604467716e-05,
+      "loss": 0.6425,
+      "step": 4632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122462568914305e-05,
+      "loss": 0.632,
+      "step": 4633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122003418782253e-05,
+      "loss": 0.5702,
+      "step": 4634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912154415407733e-05,
+      "loss": 0.5752,
+      "step": 4635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121084774805298e-05,
+      "loss": 0.6213,
+      "step": 4636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120625280971934e-05,
+      "loss": 0.5248,
+      "step": 4637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120165672583003e-05,
+      "loss": 0.5229,
+      "step": 4638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119705949644283e-05,
+      "loss": 0.586,
+      "step": 4639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119246112161542e-05,
+      "loss": 0.5739,
+      "step": 4640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911878616014056e-05,
+      "loss": 0.5984,
+      "step": 4641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118326093587116e-05,
+      "loss": 0.6559,
+      "step": 4642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117865912506984e-05,
+      "loss": 0.5377,
+      "step": 4643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117405616905945e-05,
+      "loss": 0.5707,
+      "step": 4644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116945206789782e-05,
+      "loss": 0.6433,
+      "step": 4645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116484682164274e-05,
+      "loss": 0.5954,
+      "step": 4646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116024043035213e-05,
+      "loss": 0.6119,
+      "step": 4647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115563289408375e-05,
+      "loss": 0.5353,
+      "step": 4648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911510242128955e-05,
+      "loss": 0.5838,
+      "step": 4649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114641438684534e-05,
+      "loss": 0.6176,
+      "step": 4650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114180341599107e-05,
+      "loss": 0.6119,
+      "step": 4651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113719130039065e-05,
+      "loss": 0.5758,
+      "step": 4652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91132578040102e-05,
+      "loss": 0.5897,
+      "step": 4653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112796363518308e-05,
+      "loss": 0.6061,
+      "step": 4654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911233480856918e-05,
+      "loss": 0.5543,
+      "step": 4655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911187313916862e-05,
+      "loss": 0.5125,
+      "step": 4656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111411355322422e-05,
+      "loss": 0.5799,
+      "step": 4657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110949457036387e-05,
+      "loss": 0.5861,
+      "step": 4658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110487444316317e-05,
+      "loss": 0.5739,
+      "step": 4659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110025317168014e-05,
+      "loss": 0.578,
+      "step": 4660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910956307559728e-05,
+      "loss": 0.5915,
+      "step": 4661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910910071960993e-05,
+      "loss": 0.6066,
+      "step": 4662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910863824921176e-05,
+      "loss": 0.4806,
+      "step": 4663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108175664408584e-05,
+      "loss": 0.5459,
+      "step": 4664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910771296520621e-05,
+      "loss": 0.6179,
+      "step": 4665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107250151610457e-05,
+      "loss": 0.5707,
+      "step": 4666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910678722362713e-05,
+      "loss": 0.5646,
+      "step": 4667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910632418126204e-05,
+      "loss": 0.5948,
+      "step": 4668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910586102452101e-05,
+      "loss": 0.6132,
+      "step": 4669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910539775340986e-05,
+      "loss": 0.591,
+      "step": 4670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91049343679344e-05,
+      "loss": 0.5789,
+      "step": 4671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104470868100457e-05,
+      "loss": 0.542,
+      "step": 4672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910400725391385e-05,
+      "loss": 0.5974,
+      "step": 4673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.91035435253804e-05,
+      "loss": 0.5398,
+      "step": 4674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103079682505938e-05,
+      "loss": 0.6304,
+      "step": 4675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910261572529628e-05,
+      "loss": 0.6008,
+      "step": 4676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910215165375726e-05,
+      "loss": 0.5658,
+      "step": 4677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101687467894706e-05,
+      "loss": 0.58,
+      "step": 4678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101223167714447e-05,
+      "loss": 0.6197,
+      "step": 4679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100758753222317e-05,
+      "loss": 0.6294,
+      "step": 4680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910029422442415e-05,
+      "loss": 0.5792,
+      "step": 4681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909982958132577e-05,
+      "loss": 0.5699,
+      "step": 4682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099364823933024e-05,
+      "loss": 0.5929,
+      "step": 4683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909889995225175e-05,
+      "loss": 0.5755,
+      "step": 4684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098434966287776e-05,
+      "loss": 0.57,
+      "step": 4685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097969866046957e-05,
+      "loss": 0.638,
+      "step": 4686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097504651535123e-05,
+      "loss": 0.5106,
+      "step": 4687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909703932275812e-05,
+      "loss": 0.6083,
+      "step": 4688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096573879721797e-05,
+      "loss": 0.5557,
+      "step": 4689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096108322431996e-05,
+      "loss": 0.6154,
+      "step": 4690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095642650894563e-05,
+      "loss": 0.5019,
+      "step": 4691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909517686511535e-05,
+      "loss": 0.5941,
+      "step": 4692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094710965100205e-05,
+      "loss": 0.518,
+      "step": 4693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909424495085498e-05,
+      "loss": 0.5751,
+      "step": 4694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093778822385532e-05,
+      "loss": 0.6108,
+      "step": 4695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093312579697713e-05,
+      "loss": 0.5293,
+      "step": 4696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092846222797377e-05,
+      "loss": 0.5897,
+      "step": 4697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092379751690383e-05,
+      "loss": 0.5176,
+      "step": 4698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909191316638259e-05,
+      "loss": 0.584,
+      "step": 4699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909144646687986e-05,
+      "loss": 0.5519,
+      "step": 4700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909097965318805e-05,
+      "loss": 0.5995,
+      "step": 4701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090512725313028e-05,
+      "loss": 0.6104,
+      "step": 4702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909004568326066e-05,
+      "loss": 0.6725,
+      "step": 4703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089578527036806e-05,
+      "loss": 0.4993,
+      "step": 4704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089111256647337e-05,
+      "loss": 0.5931,
+      "step": 4705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088643872098122e-05,
+      "loss": 0.6057,
+      "step": 4706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908817637339503e-05,
+      "loss": 0.5925,
+      "step": 4707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087708760543936e-05,
+      "loss": 0.5768,
+      "step": 4708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908724103355071e-05,
+      "loss": 0.5753,
+      "step": 4709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086773192421228e-05,
+      "loss": 0.6186,
+      "step": 4710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086305237161366e-05,
+      "loss": 0.6472,
+      "step": 4711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085837167776997e-05,
+      "loss": 0.568,
+      "step": 4712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085368984274007e-05,
+      "loss": 0.5822,
+      "step": 4713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084900686658278e-05,
+      "loss": 0.511,
+      "step": 4714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084432274935684e-05,
+      "loss": 0.6241,
+      "step": 4715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083963749112108e-05,
+      "loss": 0.5854,
+      "step": 4716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083495109193444e-05,
+      "loss": 0.6364,
+      "step": 4717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908302635518557e-05,
+      "loss": 0.5622,
+      "step": 4718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908255748709438e-05,
+      "loss": 0.5618,
+      "step": 4719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082088504925754e-05,
+      "loss": 0.6138,
+      "step": 4720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081619408685593e-05,
+      "loss": 0.5319,
+      "step": 4721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908115019837978e-05,
+      "loss": 0.6169,
+      "step": 4722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080680874014213e-05,
+      "loss": 0.6139,
+      "step": 4723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080211435594785e-05,
+      "loss": 0.5806,
+      "step": 4724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907974188312739e-05,
+      "loss": 0.6554,
+      "step": 4725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079272216617935e-05,
+      "loss": 0.5909,
+      "step": 4726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907880243607231e-05,
+      "loss": 0.5807,
+      "step": 4727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078332541496417e-05,
+      "loss": 0.5953,
+      "step": 4728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077862532896157e-05,
+      "loss": 0.5766,
+      "step": 4729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907739241027744e-05,
+      "loss": 0.5552,
+      "step": 4730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076922173646163e-05,
+      "loss": 0.5752,
+      "step": 4731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076451823008235e-05,
+      "loss": 0.5174,
+      "step": 4732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075981358369563e-05,
+      "loss": 0.6186,
+      "step": 4733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907551077973606e-05,
+      "loss": 0.5793,
+      "step": 4734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075040087113628e-05,
+      "loss": 0.5981,
+      "step": 4735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074569280508188e-05,
+      "loss": 0.6079,
+      "step": 4736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907409835992565e-05,
+      "loss": 0.5909,
+      "step": 4737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073627325371925e-05,
+      "loss": 0.5734,
+      "step": 4738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073156176852936e-05,
+      "loss": 0.5723,
+      "step": 4739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072684914374598e-05,
+      "loss": 0.5795,
+      "step": 4740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072213537942828e-05,
+      "loss": 0.5601,
+      "step": 4741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071742047563547e-05,
+      "loss": 0.6014,
+      "step": 4742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071270443242678e-05,
+      "loss": 0.5759,
+      "step": 4743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070798724986143e-05,
+      "loss": 0.6211,
+      "step": 4744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907032689279987e-05,
+      "loss": 0.5467,
+      "step": 4745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069854946689785e-05,
+      "loss": 0.5821,
+      "step": 4746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069382886661812e-05,
+      "loss": 0.5936,
+      "step": 4747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906891071272188e-05,
+      "loss": 0.5357,
+      "step": 4748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906843842487592e-05,
+      "loss": 0.5311,
+      "step": 4749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067966023129872e-05,
+      "loss": 0.5582,
+      "step": 4750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067493507489657e-05,
+      "loss": 0.6121,
+      "step": 4751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906702087796122e-05,
+      "loss": 0.5973,
+      "step": 4752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906654813455049e-05,
+      "loss": 0.5827,
+      "step": 4753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906607527726341e-05,
+      "loss": 0.6095,
+      "step": 4754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065602306105914e-05,
+      "loss": 0.5837,
+      "step": 4755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906512922108395e-05,
+      "loss": 0.5663,
+      "step": 4756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906465602220345e-05,
+      "loss": 0.5856,
+      "step": 4757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064182709470365e-05,
+      "loss": 0.63,
+      "step": 4758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906370928289064e-05,
+      "loss": 0.5738,
+      "step": 4759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063235742470213e-05,
+      "loss": 0.5988,
+      "step": 4760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062762088215042e-05,
+      "loss": 0.5625,
+      "step": 4761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906228832013107e-05,
+      "loss": 0.6071,
+      "step": 4762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061814438224248e-05,
+      "loss": 0.5863,
+      "step": 4763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061340442500535e-05,
+      "loss": 0.5706,
+      "step": 4764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906086633296587e-05,
+      "loss": 0.6394,
+      "step": 4765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906039210962622e-05,
+      "loss": 0.5651,
+      "step": 4766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.905991777248754e-05,
+      "loss": 0.5763,
+      "step": 4767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.905944332155578e-05,
+      "loss": 0.6134,
+      "step": 4768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058968756836912e-05,
+      "loss": 0.5716,
+      "step": 4769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905849407833688e-05,
+      "loss": 0.5799,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058019286061662e-05,
+      "loss": 0.6142,
+      "step": 4771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057544380017214e-05,
+      "loss": 0.5877,
+      "step": 4772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90570693602095e-05,
+      "loss": 0.5784,
+      "step": 4773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905659422664449e-05,
+      "loss": 0.5655,
+      "step": 4774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056118979328145e-05,
+      "loss": 0.548,
+      "step": 4775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905564361826644e-05,
+      "loss": 0.5329,
+      "step": 4776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055168143465344e-05,
+      "loss": 0.5247,
+      "step": 4777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905469255493083e-05,
+      "loss": 0.5312,
+      "step": 4778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905421685266887e-05,
+      "loss": 0.6316,
+      "step": 4779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053741036685436e-05,
+      "loss": 0.5816,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053265106986513e-05,
+      "loss": 0.5963,
+      "step": 4781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052789063578068e-05,
+      "loss": 0.5735,
+      "step": 4782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052312906466088e-05,
+      "loss": 0.6121,
+      "step": 4783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905183663565655e-05,
+      "loss": 0.5383,
+      "step": 4784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905136025115544e-05,
+      "loss": 0.5495,
+      "step": 4785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050883752968733e-05,
+      "loss": 0.6324,
+      "step": 4786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050407141102423e-05,
+      "loss": 0.5716,
+      "step": 4787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049930415562493e-05,
+      "loss": 0.62,
+      "step": 4788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904945357635493e-05,
+      "loss": 0.6384,
+      "step": 4789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904897662348572e-05,
+      "loss": 0.6265,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904849955696086e-05,
+      "loss": 0.5883,
+      "step": 4791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048022376786335e-05,
+      "loss": 0.5295,
+      "step": 4792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047545082968148e-05,
+      "loss": 0.5702,
+      "step": 4793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047067675512285e-05,
+      "loss": 0.6044,
+      "step": 4794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046590154424743e-05,
+      "loss": 0.5159,
+      "step": 4795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046112519711525e-05,
+      "loss": 0.6014,
+      "step": 4796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045634771378627e-05,
+      "loss": 0.5995,
+      "step": 4797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045156909432047e-05,
+      "loss": 0.6201,
+      "step": 4798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044678933877793e-05,
+      "loss": 0.573,
+      "step": 4799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904420084472186e-05,
+      "loss": 0.5899,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043722641970264e-05,
+      "loss": 0.6462,
+      "step": 4801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043244325629004e-05,
+      "loss": 0.4842,
+      "step": 4802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042765895704085e-05,
+      "loss": 0.5797,
+      "step": 4803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042287352201524e-05,
+      "loss": 0.6311,
+      "step": 4804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041808695127325e-05,
+      "loss": 0.5056,
+      "step": 4805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041329924487503e-05,
+      "loss": 0.612,
+      "step": 4806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040851040288067e-05,
+      "loss": 0.5644,
+      "step": 4807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040372042535042e-05,
+      "loss": 0.6274,
+      "step": 4808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039892931234434e-05,
+      "loss": 0.6426,
+      "step": 4809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039413706392266e-05,
+      "loss": 0.5865,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038934368014556e-05,
+      "loss": 0.5504,
+      "step": 4811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038454916107324e-05,
+      "loss": 0.6131,
+      "step": 4812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903797535067659e-05,
+      "loss": 0.6116,
+      "step": 4813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903749567172838e-05,
+      "loss": 0.527,
+      "step": 4814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037015879268723e-05,
+      "loss": 0.5773,
+      "step": 4815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036535973303634e-05,
+      "loss": 0.562,
+      "step": 4816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903605595383915e-05,
+      "loss": 0.5628,
+      "step": 4817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035575820881296e-05,
+      "loss": 0.5758,
+      "step": 4818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035095574436105e-05,
+      "loss": 0.575,
+      "step": 4819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034615214509607e-05,
+      "loss": 0.6058,
+      "step": 4820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034134741107836e-05,
+      "loss": 0.5999,
+      "step": 4821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033654154236827e-05,
+      "loss": 0.6284,
+      "step": 4822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033173453902614e-05,
+      "loss": 0.5545,
+      "step": 4823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032692640111238e-05,
+      "loss": 0.5563,
+      "step": 4824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032211712868736e-05,
+      "loss": 0.6217,
+      "step": 4825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903173067218115e-05,
+      "loss": 0.5779,
+      "step": 4826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903124951805452e-05,
+      "loss": 0.5294,
+      "step": 4827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903076825049489e-05,
+      "loss": 0.557,
+      "step": 4828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9030286869508304e-05,
+      "loss": 0.5992,
+      "step": 4829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029805375100815e-05,
+      "loss": 0.6673,
+      "step": 4830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029323767278457e-05,
+      "loss": 0.5361,
+      "step": 4831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902884204604729e-05,
+      "loss": 0.5542,
+      "step": 4832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028360211413362e-05,
+      "loss": 0.5397,
+      "step": 4833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027878263382724e-05,
+      "loss": 0.6284,
+      "step": 4834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902739620196143e-05,
+      "loss": 0.5541,
+      "step": 4835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026914027155535e-05,
+      "loss": 0.6009,
+      "step": 4836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902643173897109e-05,
+      "loss": 0.5786,
+      "step": 4837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902594933741416e-05,
+      "loss": 0.5877,
+      "step": 4838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025466822490797e-05,
+      "loss": 0.5714,
+      "step": 4839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902498419420707e-05,
+      "loss": 0.6151,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024501452569036e-05,
+      "loss": 0.6341,
+      "step": 4841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024018597582756e-05,
+      "loss": 0.5811,
+      "step": 4842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023535629254296e-05,
+      "loss": 0.5781,
+      "step": 4843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023052547589724e-05,
+      "loss": 0.5443,
+      "step": 4844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022569352595108e-05,
+      "loss": 0.5734,
+      "step": 4845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022086044276517e-05,
+      "loss": 0.6165,
+      "step": 4846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021602622640018e-05,
+      "loss": 0.5399,
+      "step": 4847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021119087691683e-05,
+      "loss": 0.6017,
+      "step": 4848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902063543943759e-05,
+      "loss": 0.5289,
+      "step": 4849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902015167788381e-05,
+      "loss": 0.6328,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901966780303642e-05,
+      "loss": 0.5242,
+      "step": 4851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019183814901497e-05,
+      "loss": 0.5681,
+      "step": 4852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018699713485117e-05,
+      "loss": 0.6314,
+      "step": 4853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018215498793367e-05,
+      "loss": 0.6029,
+      "step": 4854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017731170832326e-05,
+      "loss": 0.5764,
+      "step": 4855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017246729608076e-05,
+      "loss": 0.6311,
+      "step": 4856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90167621751267e-05,
+      "loss": 0.5744,
+      "step": 4857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016277507394284e-05,
+      "loss": 0.6099,
+      "step": 4858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015792726416923e-05,
+      "loss": 0.5618,
+      "step": 4859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015307832200695e-05,
+      "loss": 0.5502,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90148228247517e-05,
+      "loss": 0.5476,
+      "step": 4861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014337704076025e-05,
+      "loss": 0.6312,
+      "step": 4862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9013852470179762e-05,
+      "loss": 0.5549,
+      "step": 4863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9013367123069008e-05,
+      "loss": 0.5892,
+      "step": 4864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012881662749857e-05,
+      "loss": 0.5234,
+      "step": 4865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012396089228413e-05,
+      "loss": 0.5678,
+      "step": 4866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011910402510762e-05,
+      "loss": 0.565,
+      "step": 4867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901142460260301e-05,
+      "loss": 0.573,
+      "step": 4868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901093868951127e-05,
+      "loss": 0.5752,
+      "step": 4869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010452663241627e-05,
+      "loss": 0.6505,
+      "step": 4870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009966523800197e-05,
+      "loss": 0.6203,
+      "step": 4871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009480271193085e-05,
+      "loss": 0.5869,
+      "step": 4872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008993905426394e-05,
+      "loss": 0.6051,
+      "step": 4873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008507426506234e-05,
+      "loss": 0.586,
+      "step": 4874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900802083443872e-05,
+      "loss": 0.5823,
+      "step": 4875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007534129229954e-05,
+      "loss": 0.6587,
+      "step": 4876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900704731088606e-05,
+      "loss": 0.6308,
+      "step": 4877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006560379413144e-05,
+      "loss": 0.5302,
+      "step": 4878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006073334817325e-05,
+      "loss": 0.6388,
+      "step": 4879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005586177104725e-05,
+      "loss": 0.5301,
+      "step": 4880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005098906281455e-05,
+      "loss": 0.6349,
+      "step": 4881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900461152235364e-05,
+      "loss": 0.5479,
+      "step": 4882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004124025327395e-05,
+      "loss": 0.6441,
+      "step": 4883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003636415208853e-05,
+      "loss": 0.5367,
+      "step": 4884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900314869200413e-05,
+      "loss": 0.6489,
+      "step": 4885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900266085571936e-05,
+      "loss": 0.5658,
+      "step": 4886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002172906360662e-05,
+      "loss": 0.5362,
+      "step": 4887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900168484393417e-05,
+      "loss": 0.5865,
+      "step": 4888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900119666844601e-05,
+      "loss": 0.5437,
+      "step": 4889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000708379902315e-05,
+      "loss": 0.6085,
+      "step": 4890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000219978309224e-05,
+      "loss": 0.602,
+      "step": 4891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899973146367286e-05,
+      "loss": 0.5441,
+      "step": 4892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999242835999365e-05,
+      "loss": 0.513,
+      "step": 4893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899875409529488e-05,
+      "loss": 0.5976,
+      "step": 4894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998265241565538e-05,
+      "loss": 0.5365,
+      "step": 4895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899777627481748e-05,
+      "loss": 0.5653,
+      "step": 4896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997287195056847e-05,
+      "loss": 0.6239,
+      "step": 4897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996798002289786e-05,
+      "loss": 0.6217,
+      "step": 4898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996308696522435e-05,
+      "loss": 0.6918,
+      "step": 4899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995819277760945e-05,
+      "loss": 0.5636,
+      "step": 4900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995329746011458e-05,
+      "loss": 0.5533,
+      "step": 4901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899484010128013e-05,
+      "loss": 0.5663,
+      "step": 4902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994350343573103e-05,
+      "loss": 0.6284,
+      "step": 4903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993860472896535e-05,
+      "loss": 0.5704,
+      "step": 4904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993370489256576e-05,
+      "loss": 0.4956,
+      "step": 4905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992880392659376e-05,
+      "loss": 0.6078,
+      "step": 4906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89923901831111e-05,
+      "loss": 0.5952,
+      "step": 4907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991899860617896e-05,
+      "loss": 0.5662,
+      "step": 4908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991409425185925e-05,
+      "loss": 0.5261,
+      "step": 4909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899091887682135e-05,
+      "loss": 0.6288,
+      "step": 4910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899042821553033e-05,
+      "loss": 0.6249,
+      "step": 4911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898993744131903e-05,
+      "loss": 0.6068,
+      "step": 4912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898944655419361e-05,
+      "loss": 0.5631,
+      "step": 4913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988955554160234e-05,
+      "loss": 0.664,
+      "step": 4914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988464441225077e-05,
+      "loss": 0.5205,
+      "step": 4915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987973215394303e-05,
+      "loss": 0.5712,
+      "step": 4916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987481876674085e-05,
+      "loss": 0.5163,
+      "step": 4917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986990425070585e-05,
+      "loss": 0.5761,
+      "step": 4918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986498860589987e-05,
+      "loss": 0.5764,
+      "step": 4919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898600718323846e-05,
+      "loss": 0.5649,
+      "step": 4920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985515393022177e-05,
+      "loss": 0.558,
+      "step": 4921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898502348994732e-05,
+      "loss": 0.5939,
+      "step": 4922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984531474020063e-05,
+      "loss": 0.5306,
+      "step": 4923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984039345246587e-05,
+      "loss": 0.5586,
+      "step": 4924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983547103633072e-05,
+      "loss": 0.5835,
+      "step": 4925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983054749185706e-05,
+      "loss": 0.57,
+      "step": 4926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982562281910667e-05,
+      "loss": 0.5946,
+      "step": 4927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982069701814147e-05,
+      "loss": 0.5704,
+      "step": 4928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981577008902325e-05,
+      "loss": 0.606,
+      "step": 4929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898108420318139e-05,
+      "loss": 0.6116,
+      "step": 4930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980591284657536e-05,
+      "loss": 0.6574,
+      "step": 4931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980098253336952e-05,
+      "loss": 0.6457,
+      "step": 4932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979605109225834e-05,
+      "loss": 0.5551,
+      "step": 4933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979111852330372e-05,
+      "loss": 0.5199,
+      "step": 4934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978618482656762e-05,
+      "loss": 0.6092,
+      "step": 4935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89781250002112e-05,
+      "loss": 0.651,
+      "step": 4936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977631404999887e-05,
+      "loss": 0.5061,
+      "step": 4937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897713769702902e-05,
+      "loss": 0.5978,
+      "step": 4938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89766438763048e-05,
+      "loss": 0.5525,
+      "step": 4939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897614994283343e-05,
+      "loss": 0.6183,
+      "step": 4940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975655896621114e-05,
+      "loss": 0.6464,
+      "step": 4941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897516173767406e-05,
+      "loss": 0.6404,
+      "step": 4942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897466746599847e-05,
+      "loss": 0.638,
+      "step": 4943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974173081600554e-05,
+      "loss": 0.5258,
+      "step": 4944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973678584486517e-05,
+      "loss": 0.5509,
+      "step": 4945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897318397466258e-05,
+      "loss": 0.5026,
+      "step": 4946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897268925213495e-05,
+      "loss": 0.5034,
+      "step": 4947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972194416909838e-05,
+      "loss": 0.583,
+      "step": 4948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971699468993463e-05,
+      "loss": 0.6005,
+      "step": 4949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897120440839204e-05,
+      "loss": 0.595,
+      "step": 4950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970709235111785e-05,
+      "loss": 0.5596,
+      "step": 4951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970213949158923e-05,
+      "loss": 0.5763,
+      "step": 4952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896971855053967e-05,
+      "loss": 0.6144,
+      "step": 4953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896922303926025e-05,
+      "loss": 0.514,
+      "step": 4954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.5934,
+      "step": 4955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89682316787458e-05,
+      "loss": 0.5461,
+      "step": 4956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967735829523225e-05,
+      "loss": 0.6172,
+      "step": 4957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967239867665386e-05,
+      "loss": 0.5521,
+      "step": 4958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966743793178515e-05,
+      "loss": 0.6382,
+      "step": 4959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966247606068835e-05,
+      "loss": 0.5947,
+      "step": 4960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965751306342584e-05,
+      "loss": 0.6217,
+      "step": 4961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965254894005993e-05,
+      "loss": 0.5398,
+      "step": 4962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964758369065303e-05,
+      "loss": 0.6131,
+      "step": 4963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896426173152674e-05,
+      "loss": 0.575,
+      "step": 4964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896376498139655e-05,
+      "loss": 0.5639,
+      "step": 4965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963268118680972e-05,
+      "loss": 0.5903,
+      "step": 4966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962771143386244e-05,
+      "loss": 0.5507,
+      "step": 4967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962274055518607e-05,
+      "loss": 0.5936,
+      "step": 4968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961776855084307e-05,
+      "loss": 0.4689,
+      "step": 4969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961279542089587e-05,
+      "loss": 0.5247,
+      "step": 4970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960782116540692e-05,
+      "loss": 0.6225,
+      "step": 4971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960284578443873e-05,
+      "loss": 0.5567,
+      "step": 4972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895978692780538e-05,
+      "loss": 0.5805,
+      "step": 4973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895928916463146e-05,
+      "loss": 0.4787,
+      "step": 4974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895879128892837e-05,
+      "loss": 0.6081,
+      "step": 4975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958293300702353e-05,
+      "loss": 0.5482,
+      "step": 4976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957795199959674e-05,
+      "loss": 0.5985,
+      "step": 4977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957296986706584e-05,
+      "loss": 0.5839,
+      "step": 4978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956798660949345e-05,
+      "loss": 0.6184,
+      "step": 4979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895630022269421e-05,
+      "loss": 0.6067,
+      "step": 4980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955801671947448e-05,
+      "loss": 0.5587,
+      "step": 4981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955303008715308e-05,
+      "loss": 0.5892,
+      "step": 4982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954804233004064e-05,
+      "loss": 0.6047,
+      "step": 4983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895430534481998e-05,
+      "loss": 0.5611,
+      "step": 4984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953806344169315e-05,
+      "loss": 0.6875,
+      "step": 4985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953307231058343e-05,
+      "loss": 0.5253,
+      "step": 4986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895280800549333e-05,
+      "loss": 0.617,
+      "step": 4987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952308667480546e-05,
+      "loss": 0.5991,
+      "step": 4988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895180921702627e-05,
+      "loss": 0.5526,
+      "step": 4989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951309654136763e-05,
+      "loss": 0.5815,
+      "step": 4990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950809978818304e-05,
+      "loss": 0.5512,
+      "step": 4991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950310191077173e-05,
+      "loss": 0.6383,
+      "step": 4992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949810290919646e-05,
+      "loss": 0.5237,
+      "step": 4993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949310278351997e-05,
+      "loss": 0.562,
+      "step": 4994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948810153380514e-05,
+      "loss": 0.6162,
+      "step": 4995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948309916011473e-05,
+      "loss": 0.5979,
+      "step": 4996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947809566251157e-05,
+      "loss": 0.5179,
+      "step": 4997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894730910410585e-05,
+      "loss": 0.5417,
+      "step": 4998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946808529581843e-05,
+      "loss": 0.5176,
+      "step": 4999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946307842685416e-05,
+      "loss": 0.6013,
+      "step": 5000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945807043422863e-05,
+      "loss": 0.5617,
+      "step": 5001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945306131800474e-05,
+      "loss": 0.5322,
+      "step": 5002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894480510782454e-05,
+      "loss": 0.54,
+      "step": 5003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894430397150135e-05,
+      "loss": 0.5513,
+      "step": 5004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943802722837204e-05,
+      "loss": 0.5587,
+      "step": 5005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943301361838392e-05,
+      "loss": 0.505,
+      "step": 5006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942799888511217e-05,
+      "loss": 0.6072,
+      "step": 5007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894229830286197e-05,
+      "loss": 0.5785,
+      "step": 5008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894179660489696e-05,
+      "loss": 0.6002,
+      "step": 5009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941294794622483e-05,
+      "loss": 0.5696,
+      "step": 5010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894079287204484e-05,
+      "loss": 0.6225,
+      "step": 5011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894029083717034e-05,
+      "loss": 0.6061,
+      "step": 5012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893978869000529e-05,
+      "loss": 0.6171,
+      "step": 5013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939286430555987e-05,
+      "loss": 0.5521,
+      "step": 5014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938784058828747e-05,
+      "loss": 0.5664,
+      "step": 5015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893828157482988e-05,
+      "loss": 0.5874,
+      "step": 5016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937778978565695e-05,
+      "loss": 0.5423,
+      "step": 5017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937276270042506e-05,
+      "loss": 0.4569,
+      "step": 5018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893677344926663e-05,
+      "loss": 0.5718,
+      "step": 5019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936270516244374e-05,
+      "loss": 0.5942,
+      "step": 5020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893576747098206e-05,
+      "loss": 0.521,
+      "step": 5021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935264313486006e-05,
+      "loss": 0.462,
+      "step": 5022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934761043762534e-05,
+      "loss": 0.6051,
+      "step": 5023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893425766181796e-05,
+      "loss": 0.5407,
+      "step": 5024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933754167658614e-05,
+      "loss": 0.6114,
+      "step": 5025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893325056129081e-05,
+      "loss": 0.5707,
+      "step": 5026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932746842720884e-05,
+      "loss": 0.6214,
+      "step": 5027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932243011955154e-05,
+      "loss": 0.587,
+      "step": 5028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893173906899995e-05,
+      "loss": 0.5685,
+      "step": 5029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931235013861608e-05,
+      "loss": 0.5212,
+      "step": 5030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930730846546452e-05,
+      "loss": 0.5822,
+      "step": 5031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930226567060814e-05,
+      "loss": 0.5393,
+      "step": 5032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929722175411033e-05,
+      "loss": 0.5772,
+      "step": 5033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892921767160344e-05,
+      "loss": 0.6001,
+      "step": 5034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928713055644375e-05,
+      "loss": 0.5071,
+      "step": 5035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892820832754017e-05,
+      "loss": 0.6251,
+      "step": 5036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927703487297174e-05,
+      "loss": 0.5655,
+      "step": 5037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927198534921718e-05,
+      "loss": 0.5987,
+      "step": 5038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926693470420148e-05,
+      "loss": 0.5598,
+      "step": 5039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892618829379881e-05,
+      "loss": 0.5864,
+      "step": 5040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925683005064045e-05,
+      "loss": 0.6227,
+      "step": 5041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89251776042222e-05,
+      "loss": 0.5965,
+      "step": 5042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892467209127963e-05,
+      "loss": 0.567,
+      "step": 5043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924166466242677e-05,
+      "loss": 0.5472,
+      "step": 5044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892366072911769e-05,
+      "loss": 0.5698,
+      "step": 5045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892315487991103e-05,
+      "loss": 0.6359,
+      "step": 5046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892264891862904e-05,
+      "loss": 0.5465,
+      "step": 5047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922142845278083e-05,
+      "loss": 0.5812,
+      "step": 5048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892163665986451e-05,
+      "loss": 0.5697,
+      "step": 5049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892113036239468e-05,
+      "loss": 0.5835,
+      "step": 5050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920623952874958e-05,
+      "loss": 0.6382,
+      "step": 5051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920117431311696e-05,
+      "loss": 0.5841,
+      "step": 5052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919610797711257e-05,
+      "loss": 0.5994,
+      "step": 5053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891910405208001e-05,
+      "loss": 0.5879,
+      "step": 5054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918597194424318e-05,
+      "loss": 0.5122,
+      "step": 5055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918090224750543e-05,
+      "loss": 0.6399,
+      "step": 5056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8917583143065058e-05,
+      "loss": 0.6367,
+      "step": 5057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917075949374226e-05,
+      "loss": 0.582,
+      "step": 5058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891656864368442e-05,
+      "loss": 0.528,
+      "step": 5059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916061226002017e-05,
+      "loss": 0.6184,
+      "step": 5060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915553696333382e-05,
+      "loss": 0.5846,
+      "step": 5061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915046054684894e-05,
+      "loss": 0.5639,
+      "step": 5062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891453830106293e-05,
+      "loss": 0.6237,
+      "step": 5063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914030435473866e-05,
+      "loss": 0.5908,
+      "step": 5064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913522457924078e-05,
+      "loss": 0.5592,
+      "step": 5065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891301436841995e-05,
+      "loss": 0.5761,
+      "step": 5066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891250616696786e-05,
+      "loss": 0.5487,
+      "step": 5067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911997853574196e-05,
+      "loss": 0.5916,
+      "step": 5068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891148942824534e-05,
+      "loss": 0.6082,
+      "step": 5069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910980890987677e-05,
+      "loss": 0.5644,
+      "step": 5070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910472241807593e-05,
+      "loss": 0.5855,
+      "step": 5071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890996348071148e-05,
+      "loss": 0.6293,
+      "step": 5072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890945460770573e-05,
+      "loss": 0.604,
+      "step": 5073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908945622796727e-05,
+      "loss": 0.5518,
+      "step": 5074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890843652599087e-05,
+      "loss": 0.6041,
+      "step": 5075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890792731729455e-05,
+      "loss": 0.5897,
+      "step": 5076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907417996714167e-05,
+      "loss": 0.5319,
+      "step": 5077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890690856425611e-05,
+      "loss": 0.5899,
+      "step": 5078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906399019926785e-05,
+      "loss": 0.5896,
+      "step": 5079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890588936373259e-05,
+      "loss": 0.625,
+      "step": 5080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905379595679924e-05,
+      "loss": 0.6067,
+      "step": 5081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904869715775195e-05,
+      "loss": 0.5968,
+      "step": 5082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904359724024803e-05,
+      "loss": 0.6284,
+      "step": 5083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890384962043515e-05,
+      "loss": 0.5282,
+      "step": 5084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903339405012652e-05,
+      "loss": 0.5736,
+      "step": 5085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890282907776371e-05,
+      "loss": 0.579,
+      "step": 5086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902318638694738e-05,
+      "loss": 0.5414,
+      "step": 5087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890180808781214e-05,
+      "loss": 0.513,
+      "step": 5088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890129742512234e-05,
+      "loss": 0.5361,
+      "step": 5089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900786650631745e-05,
+      "loss": 0.5582,
+      "step": 5090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890027576434677e-05,
+      "loss": 0.5515,
+      "step": 5091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889976476627383e-05,
+      "loss": 0.5635,
+      "step": 5092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899253656419348e-05,
+      "loss": 0.5857,
+      "step": 5093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898742434789744e-05,
+      "loss": 0.5959,
+      "step": 5094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898231101391436e-05,
+      "loss": 0.592,
+      "step": 5095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897719656230843e-05,
+      "loss": 0.6461,
+      "step": 5096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88972080993144e-05,
+      "loss": 0.6136,
+      "step": 5097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889669643064852e-05,
+      "loss": 0.6039,
+      "step": 5098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896184650239635e-05,
+      "loss": 0.5377,
+      "step": 5099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895672758094172e-05,
+      "loss": 0.5691,
+      "step": 5100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895160754218562e-05,
+      "loss": 0.576,
+      "step": 5101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894648638619235e-05,
+      "loss": 0.6474,
+      "step": 5102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894136411302622e-05,
+      "loss": 0.5887,
+      "step": 5103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893624072275157e-05,
+      "loss": 0.5418,
+      "step": 5104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893111621543277e-05,
+      "loss": 0.5874,
+      "step": 5105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892599059113416e-05,
+      "loss": 0.5807,
+      "step": 5106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892086384992016e-05,
+      "loss": 0.5154,
+      "step": 5107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889157359918551e-05,
+      "loss": 0.5547,
+      "step": 5108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891060701700342e-05,
+      "loss": 0.6268,
+      "step": 5109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890547692542954e-05,
+      "loss": 0.5552,
+      "step": 5110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889003457171979e-05,
+      "loss": 0.5737,
+      "step": 5111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888952133923729e-05,
+      "loss": 0.5763,
+      "step": 5112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889007995101907e-05,
+      "loss": 0.5857,
+      "step": 5113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888494539320085e-05,
+      "loss": 0.5324,
+      "step": 5114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887980971898273e-05,
+      "loss": 0.526,
+      "step": 5115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887467292842924e-05,
+      "loss": 0.5705,
+      "step": 5116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886953502160486e-05,
+      "loss": 0.6429,
+      "step": 5117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888643959985742e-05,
+      "loss": 0.489,
+      "step": 5118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888592558594017e-05,
+      "loss": 0.5716,
+      "step": 5119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885411460415195e-05,
+      "loss": 0.5579,
+      "step": 5120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888489722328896e-05,
+      "loss": 0.5557,
+      "step": 5121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884382874567914e-05,
+      "loss": 0.5612,
+      "step": 5122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883868414258525e-05,
+      "loss": 0.5406,
+      "step": 5123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883353842367247e-05,
+      "loss": 0.5606,
+      "step": 5124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888283915890055e-05,
+      "loss": 0.5998,
+      "step": 5125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882324363864893e-05,
+      "loss": 0.5987,
+      "step": 5126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881809457266746e-05,
+      "loss": 0.6195,
+      "step": 5127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881294439112575e-05,
+      "loss": 0.4672,
+      "step": 5128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888077930940885e-05,
+      "loss": 0.533,
+      "step": 5129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880264068162037e-05,
+      "loss": 0.5403,
+      "step": 5130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879748715378612e-05,
+      "loss": 0.5714,
+      "step": 5131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879233251065045e-05,
+      "loss": 0.5747,
+      "step": 5132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887871767522781e-05,
+      "loss": 0.6004,
+      "step": 5133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878201987873385e-05,
+      "loss": 0.5629,
+      "step": 5134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877686189008245e-05,
+      "loss": 0.4868,
+      "step": 5135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887717027863887e-05,
+      "loss": 0.6001,
+      "step": 5136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887665425677174e-05,
+      "loss": 0.6216,
+      "step": 5137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876138123413335e-05,
+      "loss": 0.5578,
+      "step": 5138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875621878570135e-05,
+      "loss": 0.5542,
+      "step": 5139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887510552224863e-05,
+      "loss": 0.5381,
+      "step": 5140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874589054455305e-05,
+      "loss": 0.6203,
+      "step": 5141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887407247519664e-05,
+      "loss": 0.5793,
+      "step": 5142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873555784479126e-05,
+      "loss": 0.5894,
+      "step": 5143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887303898230926e-05,
+      "loss": 0.5225,
+      "step": 5144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872522068693528e-05,
+      "loss": 0.6023,
+      "step": 5145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872005043638414e-05,
+      "loss": 0.5524,
+      "step": 5146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871487907150428e-05,
+      "loss": 0.5571,
+      "step": 5147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870970659236053e-05,
+      "loss": 0.5596,
+      "step": 5148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887045329990179e-05,
+      "loss": 0.6668,
+      "step": 5149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869935829154138e-05,
+      "loss": 0.6126,
+      "step": 5150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869418246999593e-05,
+      "loss": 0.638,
+      "step": 5151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8868900553444658e-05,
+      "loss": 0.5687,
+      "step": 5152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8868382748495837e-05,
+      "loss": 0.6605,
+      "step": 5153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867864832159632e-05,
+      "loss": 0.475,
+      "step": 5154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886734680444255e-05,
+      "loss": 0.558,
+      "step": 5155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866828665351092e-05,
+      "loss": 0.5711,
+      "step": 5156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886631041489177e-05,
+      "loss": 0.5823,
+      "step": 5157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865792053071096e-05,
+      "loss": 0.5249,
+      "step": 5158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865273579895574e-05,
+      "loss": 0.5523,
+      "step": 5159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864754995371717e-05,
+      "loss": 0.5851,
+      "step": 5160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864236299506046e-05,
+      "loss": 0.5384,
+      "step": 5161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863717492305066e-05,
+      "loss": 0.5753,
+      "step": 5162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88631985737753e-05,
+      "loss": 0.5894,
+      "step": 5163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862679543923265e-05,
+      "loss": 0.5642,
+      "step": 5164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862160402755477e-05,
+      "loss": 0.561,
+      "step": 5165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886164115027846e-05,
+      "loss": 0.5713,
+      "step": 5166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886112178649873e-05,
+      "loss": 0.5485,
+      "step": 5167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860602311422816e-05,
+      "loss": 0.5993,
+      "step": 5168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886008272505724e-05,
+      "loss": 0.6463,
+      "step": 5169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859563027408533e-05,
+      "loss": 0.5824,
+      "step": 5170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859043218483212e-05,
+      "loss": 0.6234,
+      "step": 5171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858523298287818e-05,
+      "loss": 0.6111,
+      "step": 5172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885800326682887e-05,
+      "loss": 0.5585,
+      "step": 5173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885748312411291e-05,
+      "loss": 0.5229,
+      "step": 5174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885696287014646e-05,
+      "loss": 0.5667,
+      "step": 5175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885644250493606e-05,
+      "loss": 0.5255,
+      "step": 5176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855922028488253e-05,
+      "loss": 0.5554,
+      "step": 5177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855401440809565e-05,
+      "loss": 0.6503,
+      "step": 5178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885488074190654e-05,
+      "loss": 0.6069,
+      "step": 5179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854359931785714e-05,
+      "loss": 0.6229,
+      "step": 5180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853839010453632e-05,
+      "loss": 0.5955,
+      "step": 5181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853317977916842e-05,
+      "loss": 0.5351,
+      "step": 5182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852796834181878e-05,
+      "loss": 0.6589,
+      "step": 5183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885227557925529e-05,
+      "loss": 0.4773,
+      "step": 5184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851754213143626e-05,
+      "loss": 0.5798,
+      "step": 5185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885123273585343e-05,
+      "loss": 0.6187,
+      "step": 5186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850711147391257e-05,
+      "loss": 0.62,
+      "step": 5187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850189447763657e-05,
+      "loss": 0.6433,
+      "step": 5188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849667636977186e-05,
+      "loss": 0.5895,
+      "step": 5189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884914571503839e-05,
+      "loss": 0.6494,
+      "step": 5190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848623681953826e-05,
+      "loss": 0.5805,
+      "step": 5191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848101537730053e-05,
+      "loss": 0.5781,
+      "step": 5192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847579282373634e-05,
+      "loss": 0.6299,
+      "step": 5193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884705691589112e-05,
+      "loss": 0.6234,
+      "step": 5194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846534438289077e-05,
+      "loss": 0.5809,
+      "step": 5195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846011849574065e-05,
+      "loss": 0.5884,
+      "step": 5196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845489149752648e-05,
+      "loss": 0.553,
+      "step": 5197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844966338831394e-05,
+      "loss": 0.6497,
+      "step": 5198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844443416816863e-05,
+      "loss": 0.6129,
+      "step": 5199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843920383715634e-05,
+      "loss": 0.5314,
+      "step": 5200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843397239534265e-05,
+      "loss": 0.5385,
+      "step": 5201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842873984279334e-05,
+      "loss": 0.5976,
+      "step": 5202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884235061795741e-05,
+      "loss": 0.5774,
+      "step": 5203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841827140575065e-05,
+      "loss": 0.5806,
+      "step": 5204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841303552138878e-05,
+      "loss": 0.5957,
+      "step": 5205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840779852655423e-05,
+      "loss": 0.5684,
+      "step": 5206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840256042131276e-05,
+      "loss": 0.5802,
+      "step": 5207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883973212057302e-05,
+      "loss": 0.6233,
+      "step": 5208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839208087987234e-05,
+      "loss": 0.5786,
+      "step": 5209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838683944380496e-05,
+      "loss": 0.5802,
+      "step": 5210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838159689759393e-05,
+      "loss": 0.5394,
+      "step": 5211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837635324130513e-05,
+      "loss": 0.4868,
+      "step": 5212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883711084750044e-05,
+      "loss": 0.5405,
+      "step": 5213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836586259875753e-05,
+      "loss": 0.5863,
+      "step": 5214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836061561263053e-05,
+      "loss": 0.5721,
+      "step": 5215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883553675166892e-05,
+      "loss": 0.605,
+      "step": 5216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835011831099958e-05,
+      "loss": 0.5958,
+      "step": 5217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834486799562747e-05,
+      "loss": 0.5652,
+      "step": 5218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833961657063887e-05,
+      "loss": 0.6107,
+      "step": 5219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833436403609973e-05,
+      "loss": 0.6776,
+      "step": 5220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832911039207604e-05,
+      "loss": 0.5294,
+      "step": 5221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883238556386338e-05,
+      "loss": 0.5658,
+      "step": 5222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88318599775839e-05,
+      "loss": 0.6445,
+      "step": 5223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883133428037576e-05,
+      "loss": 0.5681,
+      "step": 5224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830808472245565e-05,
+      "loss": 0.5402,
+      "step": 5225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830282553199927e-05,
+      "loss": 0.6083,
+      "step": 5226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829756523245442e-05,
+      "loss": 0.5708,
+      "step": 5227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829230382388722e-05,
+      "loss": 0.5839,
+      "step": 5228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828704130636373e-05,
+      "loss": 0.6184,
+      "step": 5229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828177767995006e-05,
+      "loss": 0.5209,
+      "step": 5230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882765129447123e-05,
+      "loss": 0.6203,
+      "step": 5231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882712471007166e-05,
+      "loss": 0.5841,
+      "step": 5232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826598014802912e-05,
+      "loss": 0.5237,
+      "step": 5233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826071208671598e-05,
+      "loss": 0.5961,
+      "step": 5234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825544291684332e-05,
+      "loss": 0.5417,
+      "step": 5235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825017263847737e-05,
+      "loss": 0.5436,
+      "step": 5236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824490125168432e-05,
+      "loss": 0.5733,
+      "step": 5237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823962875653036e-05,
+      "loss": 0.5251,
+      "step": 5238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823435515308173e-05,
+      "loss": 0.5498,
+      "step": 5239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822908044140463e-05,
+      "loss": 0.6312,
+      "step": 5240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822380462156538e-05,
+      "loss": 0.5699,
+      "step": 5241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821852769363017e-05,
+      "loss": 0.5546,
+      "step": 5242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882132496576653e-05,
+      "loss": 0.5934,
+      "step": 5243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820797051373713e-05,
+      "loss": 0.5003,
+      "step": 5244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882026902619119e-05,
+      "loss": 0.5606,
+      "step": 5245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819740890225588e-05,
+      "loss": 0.6046,
+      "step": 5246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.5376,
+      "step": 5247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881868428597171e-05,
+      "loss": 0.6078,
+      "step": 5248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88181558176967e-05,
+      "loss": 0.595,
+      "step": 5249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817627238665155e-05,
+      "loss": 0.5509,
+      "step": 5250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881709854888372e-05,
+      "loss": 0.5716,
+      "step": 5251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816569748359037e-05,
+      "loss": 0.6272,
+      "step": 5252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816040837097742e-05,
+      "loss": 0.5375,
+      "step": 5253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881551181510648e-05,
+      "loss": 0.5224,
+      "step": 5254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814982682391893e-05,
+      "loss": 0.5932,
+      "step": 5255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814453438960634e-05,
+      "loss": 0.5617,
+      "step": 5256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813924084819347e-05,
+      "loss": 0.5644,
+      "step": 5257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813394619974675e-05,
+      "loss": 0.5083,
+      "step": 5258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812865044433277e-05,
+      "loss": 0.5677,
+      "step": 5259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88123353582018e-05,
+      "loss": 0.5554,
+      "step": 5260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811805561286896e-05,
+      "loss": 0.5992,
+      "step": 5261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881127565369522e-05,
+      "loss": 0.6151,
+      "step": 5262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810745635433433e-05,
+      "loss": 0.5709,
+      "step": 5263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810215506508182e-05,
+      "loss": 0.6169,
+      "step": 5264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880968526692613e-05,
+      "loss": 0.5401,
+      "step": 5265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880915491669394e-05,
+      "loss": 0.5849,
+      "step": 5266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808624455818272e-05,
+      "loss": 0.5888,
+      "step": 5267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808093884305786e-05,
+      "loss": 0.5773,
+      "step": 5268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807563202163147e-05,
+      "loss": 0.5619,
+      "step": 5269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807032409397026e-05,
+      "loss": 0.5945,
+      "step": 5270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806501506014077e-05,
+      "loss": 0.6712,
+      "step": 5271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805970492020982e-05,
+      "loss": 0.616,
+      "step": 5272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88054393674244e-05,
+      "loss": 0.6213,
+      "step": 5273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880490813223101e-05,
+      "loss": 0.6315,
+      "step": 5274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880437678644748e-05,
+      "loss": 0.4983,
+      "step": 5275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803845330080484e-05,
+      "loss": 0.5141,
+      "step": 5276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803313763136698e-05,
+      "loss": 0.5463,
+      "step": 5277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802782085622796e-05,
+      "loss": 0.6373,
+      "step": 5278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880225029754546e-05,
+      "loss": 0.5323,
+      "step": 5279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801718398911365e-05,
+      "loss": 0.5427,
+      "step": 5280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801186389727195e-05,
+      "loss": 0.5667,
+      "step": 5281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880065426999963e-05,
+      "loss": 0.5803,
+      "step": 5282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880012203973536e-05,
+      "loss": 0.573,
+      "step": 5283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879958969894106e-05,
+      "loss": 0.608,
+      "step": 5284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799057247623418e-05,
+      "loss": 0.5501,
+      "step": 5285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798524685789128e-05,
+      "loss": 0.6121,
+      "step": 5286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797992013444873e-05,
+      "loss": 0.6114,
+      "step": 5287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879745923059735e-05,
+      "loss": 0.6572,
+      "step": 5288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879692633725324e-05,
+      "loss": 0.5882,
+      "step": 5289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796393333419245e-05,
+      "loss": 0.5094,
+      "step": 5290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879586021910206e-05,
+      "loss": 0.564,
+      "step": 5291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795326994308376e-05,
+      "loss": 0.6161,
+      "step": 5292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879479365904489e-05,
+      "loss": 0.5572,
+      "step": 5293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794260213318307e-05,
+      "loss": 0.5385,
+      "step": 5294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879372665713532e-05,
+      "loss": 0.6203,
+      "step": 5295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793192990502635e-05,
+      "loss": 0.5972,
+      "step": 5296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792659213426954e-05,
+      "loss": 0.5697,
+      "step": 5297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879212532591498e-05,
+      "loss": 0.5746,
+      "step": 5298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879159132797342e-05,
+      "loss": 0.527,
+      "step": 5299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879105721960898e-05,
+      "loss": 0.5854,
+      "step": 5300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879052300082837e-05,
+      "loss": 0.5394,
+      "step": 5301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87899886716383e-05,
+      "loss": 0.5514,
+      "step": 5302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789454232045477e-05,
+      "loss": 0.5584,
+      "step": 5303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878891968205662e-05,
+      "loss": 0.5809,
+      "step": 5304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788385021678436e-05,
+      "loss": 0.6105,
+      "step": 5305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787850250917644e-05,
+      "loss": 0.5641,
+      "step": 5306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878731536978096e-05,
+      "loss": 0.6088,
+      "step": 5307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786780378275105e-05,
+      "loss": 0.5454,
+      "step": 5308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786245276406796e-05,
+      "loss": 0.5801,
+      "step": 5309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785710064182755e-05,
+      "loss": 0.6032,
+      "step": 5310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785174741609705e-05,
+      "loss": 0.606,
+      "step": 5311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784639308694364e-05,
+      "loss": 0.5689,
+      "step": 5312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784103765443464e-05,
+      "loss": 0.5418,
+      "step": 5313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878356811186373e-05,
+      "loss": 0.5874,
+      "step": 5314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783032347961884e-05,
+      "loss": 0.606,
+      "step": 5315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782496473744664e-05,
+      "loss": 0.5336,
+      "step": 5316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781960489218795e-05,
+      "loss": 0.5781,
+      "step": 5317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878142439439101e-05,
+      "loss": 0.5987,
+      "step": 5318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780888189268044e-05,
+      "loss": 0.5722,
+      "step": 5319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780351873856627e-05,
+      "loss": 0.5823,
+      "step": 5320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779815448163503e-05,
+      "loss": 0.567,
+      "step": 5321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779278912195403e-05,
+      "loss": 0.5864,
+      "step": 5322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877874226595907e-05,
+      "loss": 0.5339,
+      "step": 5323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778205509461244e-05,
+      "loss": 0.5706,
+      "step": 5324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877766864270866e-05,
+      "loss": 0.6239,
+      "step": 5325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777131665708067e-05,
+      "loss": 0.5408,
+      "step": 5326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776594578466212e-05,
+      "loss": 0.5195,
+      "step": 5327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776057380989836e-05,
+      "loss": 0.5351,
+      "step": 5328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877552007328569e-05,
+      "loss": 0.5817,
+      "step": 5329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774982655360518e-05,
+      "loss": 0.4796,
+      "step": 5330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877444512722107e-05,
+      "loss": 0.6084,
+      "step": 5331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87739074888741e-05,
+      "loss": 0.5574,
+      "step": 5332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773369740326366e-05,
+      "loss": 0.5553,
+      "step": 5333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772831881584613e-05,
+      "loss": 0.586,
+      "step": 5334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87722939126556e-05,
+      "loss": 0.5056,
+      "step": 5335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771755833546084e-05,
+      "loss": 0.5084,
+      "step": 5336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877121764426283e-05,
+      "loss": 0.6261,
+      "step": 5337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877067934481258e-05,
+      "loss": 0.5714,
+      "step": 5338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770140935202112e-05,
+      "loss": 0.61,
+      "step": 5339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769602415438184e-05,
+      "loss": 0.5598,
+      "step": 5340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769063785527556e-05,
+      "loss": 0.6373,
+      "step": 5341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768525045477e-05,
+      "loss": 0.5854,
+      "step": 5342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876798619529327e-05,
+      "loss": 0.5346,
+      "step": 5343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876744723498315e-05,
+      "loss": 0.5173,
+      "step": 5344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87669081645534e-05,
+      "loss": 0.5108,
+      "step": 5345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876636898401079e-05,
+      "loss": 0.6166,
+      "step": 5346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765829693362097e-05,
+      "loss": 0.5584,
+      "step": 5347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765290292614092e-05,
+      "loss": 0.5488,
+      "step": 5348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764750781773545e-05,
+      "loss": 0.5603,
+      "step": 5349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764211160847243e-05,
+      "loss": 0.4965,
+      "step": 5350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876367142984196e-05,
+      "loss": 0.6734,
+      "step": 5351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763131588764465e-05,
+      "loss": 0.6023,
+      "step": 5352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762591637621552e-05,
+      "loss": 0.5477,
+      "step": 5353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876205157642e-05,
+      "loss": 0.5332,
+      "step": 5354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761511405166582e-05,
+      "loss": 0.5312,
+      "step": 5355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760971123868095e-05,
+      "loss": 0.5286,
+      "step": 5356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760430732531318e-05,
+      "loss": 0.5837,
+      "step": 5357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875989023116304e-05,
+      "loss": 0.5796,
+      "step": 5358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759349619770053e-05,
+      "loss": 0.604,
+      "step": 5359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758808898359142e-05,
+      "loss": 0.5672,
+      "step": 5360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758268066937103e-05,
+      "loss": 0.514,
+      "step": 5361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757727125510722e-05,
+      "loss": 0.6705,
+      "step": 5362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87571860740868e-05,
+      "loss": 0.5063,
+      "step": 5363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756644912672132e-05,
+      "loss": 0.6486,
+      "step": 5364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875610364127351e-05,
+      "loss": 0.5262,
+      "step": 5365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755562259897738e-05,
+      "loss": 0.5762,
+      "step": 5366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755020768551615e-05,
+      "loss": 0.5145,
+      "step": 5367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875447916724194e-05,
+      "loss": 0.543,
+      "step": 5368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753937455975516e-05,
+      "loss": 0.5634,
+      "step": 5369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753395634759147e-05,
+      "loss": 0.6338,
+      "step": 5370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875285370359964e-05,
+      "loss": 0.5326,
+      "step": 5371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87523116625038e-05,
+      "loss": 0.5751,
+      "step": 5372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751769511478434e-05,
+      "loss": 0.5384,
+      "step": 5373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875122725053035e-05,
+      "loss": 0.6028,
+      "step": 5374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750684879666368e-05,
+      "loss": 0.5954,
+      "step": 5375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875014239889329e-05,
+      "loss": 0.5898,
+      "step": 5376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874959980821793e-05,
+      "loss": 0.5682,
+      "step": 5377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749057107647116e-05,
+      "loss": 0.4968,
+      "step": 5378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874851429718765e-05,
+      "loss": 0.5684,
+      "step": 5379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747971376846352e-05,
+      "loss": 0.5973,
+      "step": 5380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747428346630046e-05,
+      "loss": 0.6205,
+      "step": 5381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874688520654555e-05,
+      "loss": 0.5565,
+      "step": 5382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746341956599684e-05,
+      "loss": 0.5922,
+      "step": 5383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745798596799276e-05,
+      "loss": 0.6092,
+      "step": 5384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745255127151143e-05,
+      "loss": 0.5611,
+      "step": 5385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8744711547662122e-05,
+      "loss": 0.5405,
+      "step": 5386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874416785833903e-05,
+      "loss": 0.492,
+      "step": 5387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87436240591887e-05,
+      "loss": 0.5782,
+      "step": 5388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743080150217965e-05,
+      "loss": 0.5981,
+      "step": 5389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874253613143365e-05,
+      "loss": 0.6192,
+      "step": 5390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741992002842594e-05,
+      "loss": 0.501,
+      "step": 5391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741447764451628e-05,
+      "loss": 0.6346,
+      "step": 5392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874090341626759e-05,
+      "loss": 0.5727,
+      "step": 5393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740358958297314e-05,
+      "loss": 0.6092,
+      "step": 5394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873981439054764e-05,
+      "loss": 0.5509,
+      "step": 5395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739269713025407e-05,
+      "loss": 0.5926,
+      "step": 5396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873872492573746e-05,
+      "loss": 0.5649,
+      "step": 5397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738180028690637e-05,
+      "loss": 0.5994,
+      "step": 5398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737635021891784e-05,
+      "loss": 0.6243,
+      "step": 5399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737089905347744e-05,
+      "loss": 0.528,
+      "step": 5400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736544679065365e-05,
+      "loss": 0.5163,
+      "step": 5401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87359993430515e-05,
+      "loss": 0.6229,
+      "step": 5402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873545389731299e-05,
+      "loss": 0.6018,
+      "step": 5403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873490834185669e-05,
+      "loss": 0.5988,
+      "step": 5404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734362676689452e-05,
+      "loss": 0.5953,
+      "step": 5405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873381690181813e-05,
+      "loss": 0.5638,
+      "step": 5406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873327101724958e-05,
+      "loss": 0.5802,
+      "step": 5407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732725022990656e-05,
+      "loss": 0.6528,
+      "step": 5408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732178919048217e-05,
+      "loss": 0.5057,
+      "step": 5409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873163270542912e-05,
+      "loss": 0.5004,
+      "step": 5410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731086382140228e-05,
+      "loss": 0.6403,
+      "step": 5411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87305399491884e-05,
+      "loss": 0.6197,
+      "step": 5412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729993406580503e-05,
+      "loss": 0.6086,
+      "step": 5413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729446754323405e-05,
+      "loss": 0.6233,
+      "step": 5414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728899992423957e-05,
+      "loss": 0.5241,
+      "step": 5415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728353120889044e-05,
+      "loss": 0.5863,
+      "step": 5416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727806139725525e-05,
+      "loss": 0.6016,
+      "step": 5417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872725904894027e-05,
+      "loss": 0.5237,
+      "step": 5418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726711848540152e-05,
+      "loss": 0.5401,
+      "step": 5419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872616453853205e-05,
+      "loss": 0.6128,
+      "step": 5420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872561711892283e-05,
+      "loss": 0.5139,
+      "step": 5421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725069589719367e-05,
+      "loss": 0.5919,
+      "step": 5422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724521950928546e-05,
+      "loss": 0.538,
+      "step": 5423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723974202557238e-05,
+      "loss": 0.6002,
+      "step": 5424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723426344612324e-05,
+      "loss": 0.5769,
+      "step": 5425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872287837710069e-05,
+      "loss": 0.5189,
+      "step": 5426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872233030002921e-05,
+      "loss": 0.6249,
+      "step": 5427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872178211340478e-05,
+      "loss": 0.5517,
+      "step": 5428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721233817234274e-05,
+      "loss": 0.5222,
+      "step": 5429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720685411524585e-05,
+      "loss": 0.5057,
+      "step": 5430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720136896282596e-05,
+      "loss": 0.5768,
+      "step": 5431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87195882715152e-05,
+      "loss": 0.5535,
+      "step": 5432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871903953722929e-05,
+      "loss": 0.6276,
+      "step": 5433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871849069343175e-05,
+      "loss": 0.5554,
+      "step": 5434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717941740129483e-05,
+      "loss": 0.5306,
+      "step": 5435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717392677329377e-05,
+      "loss": 0.5733,
+      "step": 5436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871684350503833e-05,
+      "loss": 0.6224,
+      "step": 5437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716294223263244e-05,
+      "loss": 0.5792,
+      "step": 5438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871574483201101e-05,
+      "loss": 0.579,
+      "step": 5439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715195331288534e-05,
+      "loss": 0.5711,
+      "step": 5440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714645721102718e-05,
+      "loss": 0.5351,
+      "step": 5441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871409600146046e-05,
+      "loss": 0.6163,
+      "step": 5442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713546172368676e-05,
+      "loss": 0.5728,
+      "step": 5443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712996233834258e-05,
+      "loss": 0.5925,
+      "step": 5444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871244618586412e-05,
+      "loss": 0.5607,
+      "step": 5445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871189602846517e-05,
+      "loss": 0.5605,
+      "step": 5446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871134576164432e-05,
+      "loss": 0.593,
+      "step": 5447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871079538540848e-05,
+      "loss": 0.5456,
+      "step": 5448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871024489976456e-05,
+      "loss": 0.5723,
+      "step": 5449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870969430471947e-05,
+      "loss": 0.6006,
+      "step": 5450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870914360028014e-05,
+      "loss": 0.526,
+      "step": 5451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870859278645348e-05,
+      "loss": 0.488,
+      "step": 5452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.87080418632464e-05,
+      "loss": 0.5399,
+      "step": 5453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870749083066583e-05,
+      "loss": 0.5049,
+      "step": 5454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706939688718687e-05,
+      "loss": 0.5589,
+      "step": 5455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706388437411895e-05,
+      "loss": 0.5605,
+      "step": 5456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705837076752377e-05,
+      "loss": 0.5707,
+      "step": 5457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705285606747055e-05,
+      "loss": 0.5839,
+      "step": 5458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870473402740286e-05,
+      "loss": 0.5024,
+      "step": 5459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870418233872672e-05,
+      "loss": 0.592,
+      "step": 5460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703630540725563e-05,
+      "loss": 0.5748,
+      "step": 5461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703078633406312e-05,
+      "loss": 0.5079,
+      "step": 5462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870252661677591e-05,
+      "loss": 0.5588,
+      "step": 5463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870197449084129e-05,
+      "loss": 0.5489,
+      "step": 5464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701422255609377e-05,
+      "loss": 0.5845,
+      "step": 5465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700869911087115e-05,
+      "loss": 0.633,
+      "step": 5466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700317457281444e-05,
+      "loss": 0.5169,
+      "step": 5467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869976489419929e-05,
+      "loss": 0.5428,
+      "step": 5468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699212221847607e-05,
+      "loss": 0.5813,
+      "step": 5469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698659440233332e-05,
+      "loss": 0.5598,
+      "step": 5470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698106549363407e-05,
+      "loss": 0.6451,
+      "step": 5471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697553549244776e-05,
+      "loss": 0.5748,
+      "step": 5472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697000439884383e-05,
+      "loss": 0.5403,
+      "step": 5473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696447221289178e-05,
+      "loss": 0.5844,
+      "step": 5474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869589389346611e-05,
+      "loss": 0.4884,
+      "step": 5475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869534045642213e-05,
+      "loss": 0.5607,
+      "step": 5476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694786910164184e-05,
+      "loss": 0.5774,
+      "step": 5477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694233254699228e-05,
+      "loss": 0.5989,
+      "step": 5478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693679490034214e-05,
+      "loss": 0.6471,
+      "step": 5479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693125616176104e-05,
+      "loss": 0.5915,
+      "step": 5480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869257163313184e-05,
+      "loss": 0.5521,
+      "step": 5481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86920175409084e-05,
+      "loss": 0.6168,
+      "step": 5482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691463339512724e-05,
+      "loss": 0.558,
+      "step": 5483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690909028951788e-05,
+      "loss": 0.5529,
+      "step": 5484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690354609232548e-05,
+      "loss": 0.5879,
+      "step": 5485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868980008036196e-05,
+      "loss": 0.587,
+      "step": 5486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689245442347002e-05,
+      "loss": 0.5523,
+      "step": 5487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688690695194633e-05,
+      "loss": 0.5622,
+      "step": 5488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688135838911826e-05,
+      "loss": 0.5717,
+      "step": 5489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868758087350554e-05,
+      "loss": 0.5211,
+      "step": 5490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687025798982758e-05,
+      "loss": 0.5652,
+      "step": 5491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868647061535044e-05,
+      "loss": 0.5635,
+      "step": 5492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685915322615566e-05,
+      "loss": 0.4875,
+      "step": 5493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685359920785112e-05,
+      "loss": 0.5524,
+      "step": 5494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868480440986605e-05,
+      "loss": 0.5887,
+      "step": 5495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684248789865356e-05,
+      "loss": 0.6071,
+      "step": 5496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683693060790013e-05,
+      "loss": 0.5618,
+      "step": 5497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683137222646996e-05,
+      "loss": 0.6553,
+      "step": 5498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868258127544329e-05,
+      "loss": 0.5936,
+      "step": 5499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868202521918588e-05,
+      "loss": 0.5478,
+      "step": 5500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681469053881744e-05,
+      "loss": 0.5049,
+      "step": 5501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868091277953787e-05,
+      "loss": 0.5734,
+      "step": 5502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680356396161248e-05,
+      "loss": 0.58,
+      "step": 5503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679799903758863e-05,
+      "loss": 0.5791,
+      "step": 5504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679243302337704e-05,
+      "loss": 0.5781,
+      "step": 5505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678686591904763e-05,
+      "loss": 0.5709,
+      "step": 5506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867812977246703e-05,
+      "loss": 0.5292,
+      "step": 5507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677572844031505e-05,
+      "loss": 0.564,
+      "step": 5508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677015806605178e-05,
+      "loss": 0.5579,
+      "step": 5509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676458660195048e-05,
+      "loss": 0.5594,
+      "step": 5510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867590140480811e-05,
+      "loss": 0.5739,
+      "step": 5511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867534404045136e-05,
+      "loss": 0.5929,
+      "step": 5512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674786567131812e-05,
+      "loss": 0.6245,
+      "step": 5513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867422898485645e-05,
+      "loss": 0.5926,
+      "step": 5514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867367129363229e-05,
+      "loss": 0.5243,
+      "step": 5515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673113493466336e-05,
+      "loss": 0.5819,
+      "step": 5516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672555584365586e-05,
+      "loss": 0.5393,
+      "step": 5517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671997566337052e-05,
+      "loss": 0.6132,
+      "step": 5518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671439439387746e-05,
+      "loss": 0.5638,
+      "step": 5519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670881203524672e-05,
+      "loss": 0.5711,
+      "step": 5520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670322858754846e-05,
+      "loss": 0.5575,
+      "step": 5521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866976440508528e-05,
+      "loss": 0.599,
+      "step": 5522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669205842522986e-05,
+      "loss": 0.5555,
+      "step": 5523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866864717107498e-05,
+      "loss": 0.5347,
+      "step": 5524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668088390748288e-05,
+      "loss": 0.6146,
+      "step": 5525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667529501549912e-05,
+      "loss": 0.5591,
+      "step": 5526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866697050348688e-05,
+      "loss": 0.5711,
+      "step": 5527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666411396566214e-05,
+      "loss": 0.6378,
+      "step": 5528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866585218079494e-05,
+      "loss": 0.4947,
+      "step": 5529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665292856180073e-05,
+      "loss": 0.596,
+      "step": 5530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664733422728646e-05,
+      "loss": 0.6393,
+      "step": 5531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664173880447678e-05,
+      "loss": 0.5514,
+      "step": 5532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663614229344203e-05,
+      "loss": 0.6457,
+      "step": 5533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663054469425244e-05,
+      "loss": 0.5551,
+      "step": 5534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866249460069784e-05,
+      "loss": 0.5781,
+      "step": 5535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866193462316902e-05,
+      "loss": 0.5408,
+      "step": 5536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661374536845812e-05,
+      "loss": 0.5203,
+      "step": 5537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8660814341735257e-05,
+      "loss": 0.6178,
+      "step": 5538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.5317,
+      "step": 5539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659693625180244e-05,
+      "loss": 0.5789,
+      "step": 5540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659133103749863e-05,
+      "loss": 0.5949,
+      "step": 5541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658572473560283e-05,
+      "loss": 0.617,
+      "step": 5542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865801173461855e-05,
+      "loss": 0.6248,
+      "step": 5543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657450886931704e-05,
+      "loss": 0.6345,
+      "step": 5544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865688993050679e-05,
+      "loss": 0.5025,
+      "step": 5545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656328865350858e-05,
+      "loss": 0.5899,
+      "step": 5546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655767691470944e-05,
+      "loss": 0.5242,
+      "step": 5547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655206408874103e-05,
+      "loss": 0.5537,
+      "step": 5548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865464501756739e-05,
+      "loss": 0.5631,
+      "step": 5549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8654083517557848e-05,
+      "loss": 0.4944,
+      "step": 5550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653521908852533e-05,
+      "loss": 0.5884,
+      "step": 5551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86529601914585e-05,
+      "loss": 0.5532,
+      "step": 5552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652398365382796e-05,
+      "loss": 0.5747,
+      "step": 5553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865183643063249e-05,
+      "loss": 0.6135,
+      "step": 5554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865127438721463e-05,
+      "loss": 0.5707,
+      "step": 5555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650712235136282e-05,
+      "loss": 0.5806,
+      "step": 5556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650149974404504e-05,
+      "loss": 0.5363,
+      "step": 5557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649587605026357e-05,
+      "loss": 0.5791,
+      "step": 5558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649025127008903e-05,
+      "loss": 0.5378,
+      "step": 5559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648462540359213e-05,
+      "loss": 0.5376,
+      "step": 5560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647899845084347e-05,
+      "loss": 0.5541,
+      "step": 5561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647337041191373e-05,
+      "loss": 0.5743,
+      "step": 5562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646774128687362e-05,
+      "loss": 0.6191,
+      "step": 5563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646211107579386e-05,
+      "loss": 0.5507,
+      "step": 5564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645647977874515e-05,
+      "loss": 0.572,
+      "step": 5565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645084739579816e-05,
+      "loss": 0.5936,
+      "step": 5566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644521392702372e-05,
+      "loss": 0.5622,
+      "step": 5567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643957937249256e-05,
+      "loss": 0.5871,
+      "step": 5568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643394373227543e-05,
+      "loss": 0.6294,
+      "step": 5569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642830700644315e-05,
+      "loss": 0.5757,
+      "step": 5570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642266919506644e-05,
+      "loss": 0.6099,
+      "step": 5571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641703029821624e-05,
+      "loss": 0.5785,
+      "step": 5572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864113903159632e-05,
+      "loss": 0.5396,
+      "step": 5573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640574924837834e-05,
+      "loss": 0.5845,
+      "step": 5574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864001070955324e-05,
+      "loss": 0.5724,
+      "step": 5575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639446385749628e-05,
+      "loss": 0.6058,
+      "step": 5576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638881953434085e-05,
+      "loss": 0.6047,
+      "step": 5577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638317412613703e-05,
+      "loss": 0.5927,
+      "step": 5578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637752763295567e-05,
+      "loss": 0.489,
+      "step": 5579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637188005486777e-05,
+      "loss": 0.57,
+      "step": 5580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863662313919442e-05,
+      "loss": 0.5153,
+      "step": 5581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636058164425586e-05,
+      "loss": 0.5546,
+      "step": 5582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863549308118739e-05,
+      "loss": 0.53,
+      "step": 5583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863492788948691e-05,
+      "loss": 0.6157,
+      "step": 5584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634362589331252e-05,
+      "loss": 0.4899,
+      "step": 5585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633797180727516e-05,
+      "loss": 0.5419,
+      "step": 5586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633231663682804e-05,
+      "loss": 0.588,
+      "step": 5587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863266603820422e-05,
+      "loss": 0.5668,
+      "step": 5588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632100304298867e-05,
+      "loss": 0.5738,
+      "step": 5589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631534461973852e-05,
+      "loss": 0.6019,
+      "step": 5590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863096851123628e-05,
+      "loss": 0.5308,
+      "step": 5591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863040245209326e-05,
+      "loss": 0.55,
+      "step": 5592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86298362845519e-05,
+      "loss": 0.584,
+      "step": 5593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629270008619315e-05,
+      "loss": 0.5943,
+      "step": 5594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628703624302617e-05,
+      "loss": 0.574,
+      "step": 5595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628137131608913e-05,
+      "loss": 0.5982,
+      "step": 5596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627570530545326e-05,
+      "loss": 0.6247,
+      "step": 5597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862700382111897e-05,
+      "loss": 0.6302,
+      "step": 5598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626437003336967e-05,
+      "loss": 0.5413,
+      "step": 5599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862587007720643e-05,
+      "loss": 0.5806,
+      "step": 5600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625303042734477e-05,
+      "loss": 0.5273,
+      "step": 5601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624735899928238e-05,
+      "loss": 0.5626,
+      "step": 5602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624168648794833e-05,
+      "loss": 0.5451,
+      "step": 5603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623601289341388e-05,
+      "loss": 0.5796,
+      "step": 5604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623033821575024e-05,
+      "loss": 0.5434,
+      "step": 5605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622466245502874e-05,
+      "loss": 0.5741,
+      "step": 5606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621898561132067e-05,
+      "loss": 0.5094,
+      "step": 5607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862133076846973e-05,
+      "loss": 0.5738,
+      "step": 5608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620762867522995e-05,
+      "loss": 0.6201,
+      "step": 5609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620194858298996e-05,
+      "loss": 0.5411,
+      "step": 5610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861962674080487e-05,
+      "loss": 0.6,
+      "step": 5611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619058515047746e-05,
+      "loss": 0.5891,
+      "step": 5612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618490181034763e-05,
+      "loss": 0.5432,
+      "step": 5613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617921738773064e-05,
+      "loss": 0.561,
+      "step": 5614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617353188269783e-05,
+      "loss": 0.5982,
+      "step": 5615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616784529532063e-05,
+      "loss": 0.5313,
+      "step": 5616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616215762567046e-05,
+      "loss": 0.612,
+      "step": 5617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861564688738188e-05,
+      "loss": 0.52,
+      "step": 5618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615077903983703e-05,
+      "loss": 0.4847,
+      "step": 5619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861450881237967e-05,
+      "loss": 0.6275,
+      "step": 5620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861393961257692e-05,
+      "loss": 0.5722,
+      "step": 5621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613370304582604e-05,
+      "loss": 0.5563,
+      "step": 5622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861280088840388e-05,
+      "loss": 0.5661,
+      "step": 5623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861223136404789e-05,
+      "loss": 0.5936,
+      "step": 5624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611661731521794e-05,
+      "loss": 0.5876,
+      "step": 5625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611091990832742e-05,
+      "loss": 0.62,
+      "step": 5626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610522141987893e-05,
+      "loss": 0.5358,
+      "step": 5627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860995218499441e-05,
+      "loss": 0.545,
+      "step": 5628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609382119859437e-05,
+      "loss": 0.6311,
+      "step": 5629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608811946590148e-05,
+      "loss": 0.5407,
+      "step": 5630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608241665193694e-05,
+      "loss": 0.5606,
+      "step": 5631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607671275677246e-05,
+      "loss": 0.5833,
+      "step": 5632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607100778047962e-05,
+      "loss": 0.5955,
+      "step": 5633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860653017231301e-05,
+      "loss": 0.5612,
+      "step": 5634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860595945847956e-05,
+      "loss": 0.5702,
+      "step": 5635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605388636554774e-05,
+      "loss": 0.5251,
+      "step": 5636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860481770654583e-05,
+      "loss": 0.5782,
+      "step": 5637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860424666845989e-05,
+      "loss": 0.566,
+      "step": 5638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603675522304128e-05,
+      "loss": 0.5896,
+      "step": 5639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603104268085724e-05,
+      "loss": 0.5596,
+      "step": 5640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602532905811848e-05,
+      "loss": 0.5901,
+      "step": 5641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601961435489673e-05,
+      "loss": 0.5763,
+      "step": 5642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601389857126383e-05,
+      "loss": 0.5642,
+      "step": 5643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600818170729156e-05,
+      "loss": 0.5392,
+      "step": 5644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860024637630517e-05,
+      "loss": 0.4981,
+      "step": 5645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599674473861608e-05,
+      "loss": 0.5703,
+      "step": 5646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599102463405652e-05,
+      "loss": 0.5572,
+      "step": 5647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598530344944488e-05,
+      "loss": 0.5673,
+      "step": 5648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85979581184853e-05,
+      "loss": 0.6089,
+      "step": 5649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859738578403528e-05,
+      "loss": 0.5719,
+      "step": 5650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596813341601608e-05,
+      "loss": 0.6256,
+      "step": 5651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596240791191483e-05,
+      "loss": 0.5311,
+      "step": 5652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859566813281209e-05,
+      "loss": 0.6513,
+      "step": 5653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595095366470625e-05,
+      "loss": 0.5701,
+      "step": 5654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594522492174284e-05,
+      "loss": 0.6023,
+      "step": 5655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593949509930252e-05,
+      "loss": 0.5619,
+      "step": 5656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593376419745735e-05,
+      "loss": 0.6089,
+      "step": 5657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8592803221627933e-05,
+      "loss": 0.5375,
+      "step": 5658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859222991558404e-05,
+      "loss": 0.5969,
+      "step": 5659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591656501621256e-05,
+      "loss": 0.5408,
+      "step": 5660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591082979746787e-05,
+      "loss": 0.6063,
+      "step": 5661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590509349967833e-05,
+      "loss": 0.5871,
+      "step": 5662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589935612291602e-05,
+      "loss": 0.6354,
+      "step": 5663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589361766725295e-05,
+      "loss": 0.5668,
+      "step": 5664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588787813276128e-05,
+      "loss": 0.6071,
+      "step": 5665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85882137519513e-05,
+      "loss": 0.5523,
+      "step": 5666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587639582758032e-05,
+      "loss": 0.6079,
+      "step": 5667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587065305703526e-05,
+      "loss": 0.5852,
+      "step": 5668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586490920795e-05,
+      "loss": 0.5306,
+      "step": 5669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858591642803967e-05,
+      "loss": 0.7194,
+      "step": 5670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585341827444746e-05,
+      "loss": 0.5193,
+      "step": 5671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584767119017445e-05,
+      "loss": 0.6087,
+      "step": 5672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858419230276499e-05,
+      "loss": 0.607,
+      "step": 5673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85836173786946e-05,
+      "loss": 0.6185,
+      "step": 5674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583042346813496e-05,
+      "loss": 0.6246,
+      "step": 5675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582467207128897e-05,
+      "loss": 0.5722,
+      "step": 5676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581891959648032e-05,
+      "loss": 0.5572,
+      "step": 5677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858131660437812e-05,
+      "loss": 0.5734,
+      "step": 5678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580741141326392e-05,
+      "loss": 0.5709,
+      "step": 5679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580165570500078e-05,
+      "loss": 0.6066,
+      "step": 5680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85795898919064e-05,
+      "loss": 0.5767,
+      "step": 5681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579014105552592e-05,
+      "loss": 0.4965,
+      "step": 5682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578438211445887e-05,
+      "loss": 0.5895,
+      "step": 5683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857786220959352e-05,
+      "loss": 0.5479,
+      "step": 5684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577286100002723e-05,
+      "loss": 0.5848,
+      "step": 5685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857670988268073e-05,
+      "loss": 0.5405,
+      "step": 5686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857613355763478e-05,
+      "loss": 0.6099,
+      "step": 5687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575557124872116e-05,
+      "loss": 0.6009,
+      "step": 5688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857498058439997e-05,
+      "loss": 0.573,
+      "step": 5689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574403936225592e-05,
+      "loss": 0.4989,
+      "step": 5690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573827180356217e-05,
+      "loss": 0.5552,
+      "step": 5691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573250316799095e-05,
+      "loss": 0.6113,
+      "step": 5692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572673345561465e-05,
+      "loss": 0.5759,
+      "step": 5693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857209626665058e-05,
+      "loss": 0.5627,
+      "step": 5694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571519080073686e-05,
+      "loss": 0.664,
+      "step": 5695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857094178583803e-05,
+      "loss": 0.529,
+      "step": 5696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857036438395087e-05,
+      "loss": 0.516,
+      "step": 5697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569786874419448e-05,
+      "loss": 0.617,
+      "step": 5698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569209257251028e-05,
+      "loss": 0.5331,
+      "step": 5699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568631532452858e-05,
+      "loss": 0.545,
+      "step": 5700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568053700032193e-05,
+      "loss": 0.5443,
+      "step": 5701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567475759996293e-05,
+      "loss": 0.5542,
+      "step": 5702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566897712352424e-05,
+      "loss": 0.5384,
+      "step": 5703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566319557107838e-05,
+      "loss": 0.6058,
+      "step": 5704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565741294269796e-05,
+      "loss": 0.5776,
+      "step": 5705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565162923845563e-05,
+      "loss": 0.5859,
+      "step": 5706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564584445842407e-05,
+      "loss": 0.5876,
+      "step": 5707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564005860267587e-05,
+      "loss": 0.6629,
+      "step": 5708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856342716712838e-05,
+      "loss": 0.5845,
+      "step": 5709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856284836643204e-05,
+      "loss": 0.589,
+      "step": 5710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856226945818585e-05,
+      "loss": 0.5808,
+      "step": 5711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561690442397077e-05,
+      "loss": 0.6155,
+      "step": 5712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856111131907299e-05,
+      "loss": 0.5628,
+      "step": 5713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560532088220867e-05,
+      "loss": 0.5161,
+      "step": 5714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855995274984798e-05,
+      "loss": 0.569,
+      "step": 5715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855937330396161e-05,
+      "loss": 0.5955,
+      "step": 5716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558793750569024e-05,
+      "loss": 0.552,
+      "step": 5717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558214089677516e-05,
+      "loss": 0.5628,
+      "step": 5718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855763432129436e-05,
+      "loss": 0.5985,
+      "step": 5719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557054445426837e-05,
+      "loss": 0.6157,
+      "step": 5720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855647446208223e-05,
+      "loss": 0.6297,
+      "step": 5721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555894371267822e-05,
+      "loss": 0.543,
+      "step": 5722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555314172990904e-05,
+      "loss": 0.599,
+      "step": 5723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554733867258763e-05,
+      "loss": 0.5409,
+      "step": 5724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554153454078684e-05,
+      "loss": 0.562,
+      "step": 5725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553572933457956e-05,
+      "loss": 0.6169,
+      "step": 5726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552992305403873e-05,
+      "loss": 0.5984,
+      "step": 5727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552411569923732e-05,
+      "loss": 0.5906,
+      "step": 5728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551830727024817e-05,
+      "loss": 0.5468,
+      "step": 5729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551249776714434e-05,
+      "loss": 0.5674,
+      "step": 5730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8550668718999873e-05,
+      "loss": 0.5426,
+      "step": 5731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550087553888436e-05,
+      "loss": 0.5963,
+      "step": 5732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549506281387417e-05,
+      "loss": 0.5638,
+      "step": 5733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548924901504125e-05,
+      "loss": 0.5906,
+      "step": 5734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854834341424585e-05,
+      "loss": 0.5487,
+      "step": 5735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854776181961991e-05,
+      "loss": 0.5873,
+      "step": 5736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547180117633597e-05,
+      "loss": 0.5257,
+      "step": 5737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854659830829423e-05,
+      "loss": 0.6155,
+      "step": 5738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546016391609104e-05,
+      "loss": 0.6124,
+      "step": 5739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545434367585536e-05,
+      "loss": 0.6296,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544852236230833e-05,
+      "loss": 0.5569,
+      "step": 5741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544269997552308e-05,
+      "loss": 0.5618,
+      "step": 5742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543687651557273e-05,
+      "loss": 0.6154,
+      "step": 5743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854310519825304e-05,
+      "loss": 0.5819,
+      "step": 5744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854252263764693e-05,
+      "loss": 0.5802,
+      "step": 5745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541939969746255e-05,
+      "loss": 0.5267,
+      "step": 5746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854135719455834e-05,
+      "loss": 0.5481,
+      "step": 5747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540774312090498e-05,
+      "loss": 0.6121,
+      "step": 5748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854019132235005e-05,
+      "loss": 0.6559,
+      "step": 5749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853960822534432e-05,
+      "loss": 0.6309,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853902502108064e-05,
+      "loss": 0.5364,
+      "step": 5751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853844170956632e-05,
+      "loss": 0.5595,
+      "step": 5752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537858290808697e-05,
+      "loss": 0.5294,
+      "step": 5753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537274764815093e-05,
+      "loss": 0.5224,
+      "step": 5754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536691131592843e-05,
+      "loss": 0.5468,
+      "step": 5755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536107391149274e-05,
+      "loss": 0.6206,
+      "step": 5756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535523543491716e-05,
+      "loss": 0.5197,
+      "step": 5757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534939588627506e-05,
+      "loss": 0.6101,
+      "step": 5758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534355526563975e-05,
+      "loss": 0.577,
+      "step": 5759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533771357308464e-05,
+      "loss": 0.5994,
+      "step": 5760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533187080868304e-05,
+      "loss": 0.5644,
+      "step": 5761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532602697250835e-05,
+      "loss": 0.5892,
+      "step": 5762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532018206463404e-05,
+      "loss": 0.5541,
+      "step": 5763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8531433608513342e-05,
+      "loss": 0.6214,
+      "step": 5764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530848903407994e-05,
+      "loss": 0.4933,
+      "step": 5765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853026409115471e-05,
+      "loss": 0.6392,
+      "step": 5766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529679171760832e-05,
+      "loss": 0.5586,
+      "step": 5767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529094145233704e-05,
+      "loss": 0.5903,
+      "step": 5768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528509011580677e-05,
+      "loss": 0.5642,
+      "step": 5769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527923770809095e-05,
+      "loss": 0.4694,
+      "step": 5770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527338422926314e-05,
+      "loss": 0.5623,
+      "step": 5771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526752967939686e-05,
+      "loss": 0.6429,
+      "step": 5772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526167405856567e-05,
+      "loss": 0.6295,
+      "step": 5773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525581736684303e-05,
+      "loss": 0.6035,
+      "step": 5774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524995960430255e-05,
+      "loss": 0.5825,
+      "step": 5775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524410077101783e-05,
+      "loss": 0.5942,
+      "step": 5776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523824086706242e-05,
+      "loss": 0.5813,
+      "step": 5777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523237989250992e-05,
+      "loss": 0.6176,
+      "step": 5778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85226517847434e-05,
+      "loss": 0.6174,
+      "step": 5779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852206547319082e-05,
+      "loss": 0.5784,
+      "step": 5780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852147905460062e-05,
+      "loss": 0.5421,
+      "step": 5781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852089252898017e-05,
+      "loss": 0.5821,
+      "step": 5782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520305896336824e-05,
+      "loss": 0.5197,
+      "step": 5783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519719156677968e-05,
+      "loss": 0.5354,
+      "step": 5784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519132310010956e-05,
+      "loss": 0.6181,
+      "step": 5785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518545356343168e-05,
+      "loss": 0.6093,
+      "step": 5786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517958295681976e-05,
+      "loss": 0.4726,
+      "step": 5787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517371128034746e-05,
+      "loss": 0.5555,
+      "step": 5788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516783853408858e-05,
+      "loss": 0.5875,
+      "step": 5789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516196471811687e-05,
+      "loss": 0.5418,
+      "step": 5790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515608983250615e-05,
+      "loss": 0.5749,
+      "step": 5791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515021387733013e-05,
+      "loss": 0.5492,
+      "step": 5792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851443368526627e-05,
+      "loss": 0.57,
+      "step": 5793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851384587585776e-05,
+      "loss": 0.5776,
+      "step": 5794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851325795951487e-05,
+      "loss": 0.5716,
+      "step": 5795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512669936244987e-05,
+      "loss": 0.6107,
+      "step": 5796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512081806055485e-05,
+      "loss": 0.6127,
+      "step": 5797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511493568953763e-05,
+      "loss": 0.5121,
+      "step": 5798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510905224947207e-05,
+      "loss": 0.5319,
+      "step": 5799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510316774043202e-05,
+      "loss": 0.5482,
+      "step": 5800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509728216249144e-05,
+      "loss": 0.5873,
+      "step": 5801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850913955157242e-05,
+      "loss": 0.5195,
+      "step": 5802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508550780020428e-05,
+      "loss": 0.5246,
+      "step": 5803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850796190160056e-05,
+      "loss": 0.5331,
+      "step": 5804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507372916320217e-05,
+      "loss": 0.5914,
+      "step": 5805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506783824186792e-05,
+      "loss": 0.5742,
+      "step": 5806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506194625207686e-05,
+      "loss": 0.6355,
+      "step": 5807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85056053193903e-05,
+      "loss": 0.5495,
+      "step": 5808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505015906742035e-05,
+      "loss": 0.5379,
+      "step": 5809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504426387270294e-05,
+      "loss": 0.5452,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850383676098248e-05,
+      "loss": 0.5302,
+      "step": 5811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503247027886e-05,
+      "loss": 0.533,
+      "step": 5812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850265718798826e-05,
+      "loss": 0.5873,
+      "step": 5813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850206724129667e-05,
+      "loss": 0.6259,
+      "step": 5814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501477187818642e-05,
+      "loss": 0.5454,
+      "step": 5815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500887027561584e-05,
+      "loss": 0.5851,
+      "step": 5816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500296760532905e-05,
+      "loss": 0.5594,
+      "step": 5817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499706386740026e-05,
+      "loss": 0.5604,
+      "step": 5818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499115906190357e-05,
+      "loss": 0.5172,
+      "step": 5819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498525318891317e-05,
+      "loss": 0.5632,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849793462485032e-05,
+      "loss": 0.5786,
+      "step": 5821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497343824074794e-05,
+      "loss": 0.5649,
+      "step": 5822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849675291657215e-05,
+      "loss": 0.5626,
+      "step": 5823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496161902349814e-05,
+      "loss": 0.6151,
+      "step": 5824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849557078141521e-05,
+      "loss": 0.6177,
+      "step": 5825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849497955377576e-05,
+      "loss": 0.5694,
+      "step": 5826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494388219438892e-05,
+      "loss": 0.5831,
+      "step": 5827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849379677841203e-05,
+      "loss": 0.5997,
+      "step": 5828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493205230702608e-05,
+      "loss": 0.5765,
+      "step": 5829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849261357631805e-05,
+      "loss": 0.6014,
+      "step": 5830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849202181526579e-05,
+      "loss": 0.5036,
+      "step": 5831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849142994755326e-05,
+      "loss": 0.6331,
+      "step": 5832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490837973187897e-05,
+      "loss": 0.608,
+      "step": 5833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849024589217713e-05,
+      "loss": 0.5709,
+      "step": 5834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489653704528404e-05,
+      "loss": 0.5326,
+      "step": 5835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489061410249148e-05,
+      "loss": 0.5608,
+      "step": 5836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488469009346805e-05,
+      "loss": 0.5691,
+      "step": 5837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8487876501828817e-05,
+      "loss": 0.5299,
+      "step": 5838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8487283887702626e-05,
+      "loss": 0.6276,
+      "step": 5839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486691166975672e-05,
+      "loss": 0.6556,
+      "step": 5840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486098339655404e-05,
+      "loss": 0.5293,
+      "step": 5841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848550540574926e-05,
+      "loss": 0.617,
+      "step": 5842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84849123652647e-05,
+      "loss": 0.6128,
+      "step": 5843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484319218209162e-05,
+      "loss": 0.5159,
+      "step": 5844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483725964590098e-05,
+      "loss": 0.5573,
+      "step": 5845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483132604414963e-05,
+      "loss": 0.6041,
+      "step": 5846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482539137691207e-05,
+      "loss": 0.6039,
+      "step": 5847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481945564426286e-05,
+      "loss": 0.597,
+      "step": 5848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481351884627652e-05,
+      "loss": 0.5335,
+      "step": 5849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480758098302764e-05,
+      "loss": 0.5244,
+      "step": 5850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480164205459078e-05,
+      "loss": 0.6247,
+      "step": 5851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479570206104054e-05,
+      "loss": 0.6717,
+      "step": 5852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478976100245155e-05,
+      "loss": 0.5526,
+      "step": 5853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847838188788984e-05,
+      "loss": 0.5594,
+      "step": 5854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477787569045573e-05,
+      "loss": 0.5537,
+      "step": 5855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847719314371982e-05,
+      "loss": 0.5753,
+      "step": 5856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847659861192005e-05,
+      "loss": 0.5633,
+      "step": 5857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476003973653723e-05,
+      "loss": 0.5565,
+      "step": 5858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475409228928314e-05,
+      "loss": 0.5348,
+      "step": 5859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847481437775129e-05,
+      "loss": 0.5105,
+      "step": 5860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847421942013012e-05,
+      "loss": 0.532,
+      "step": 5861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473624356072283e-05,
+      "loss": 0.5369,
+      "step": 5862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473029185585244e-05,
+      "loss": 0.5527,
+      "step": 5863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472433908676488e-05,
+      "loss": 0.5678,
+      "step": 5864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471838525353485e-05,
+      "loss": 0.5267,
+      "step": 5865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471243035623717e-05,
+      "loss": 0.5418,
+      "step": 5866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470647439494663e-05,
+      "loss": 0.6032,
+      "step": 5867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470051736973803e-05,
+      "loss": 0.5865,
+      "step": 5868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469455928068617e-05,
+      "loss": 0.571,
+      "step": 5869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468860012786595e-05,
+      "loss": 0.5423,
+      "step": 5870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468263991135213e-05,
+      "loss": 0.5541,
+      "step": 5871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846766786312196e-05,
+      "loss": 0.5622,
+      "step": 5872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467071628754324e-05,
+      "loss": 0.5552,
+      "step": 5873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466475288039796e-05,
+      "loss": 0.5403,
+      "step": 5874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465878840985866e-05,
+      "loss": 0.6182,
+      "step": 5875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846528228760002e-05,
+      "loss": 0.5614,
+      "step": 5876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464685627889757e-05,
+      "loss": 0.5006,
+      "step": 5877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846408886186257e-05,
+      "loss": 0.5793,
+      "step": 5878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463491989525945e-05,
+      "loss": 0.5422,
+      "step": 5879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462895010887394e-05,
+      "loss": 0.5952,
+      "step": 5880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462297925954405e-05,
+      "loss": 0.6054,
+      "step": 5881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846170073473448e-05,
+      "loss": 0.4847,
+      "step": 5882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846110343723512e-05,
+      "loss": 0.5531,
+      "step": 5883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846050603346383e-05,
+      "loss": 0.5642,
+      "step": 5884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459908523428107e-05,
+      "loss": 0.5976,
+      "step": 5885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845931090713546e-05,
+      "loss": 0.5921,
+      "step": 5886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458713184593394e-05,
+      "loss": 0.5745,
+      "step": 5887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458115355809416e-05,
+      "loss": 0.587,
+      "step": 5888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8457517420791037e-05,
+      "loss": 0.551,
+      "step": 5889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456919379545765e-05,
+      "loss": 0.5559,
+      "step": 5890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845632123208111e-05,
+      "loss": 0.5622,
+      "step": 5891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845572297840459e-05,
+      "loss": 0.6196,
+      "step": 5892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455124618523714e-05,
+      "loss": 0.513,
+      "step": 5893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454526152446e-05,
+      "loss": 0.5919,
+      "step": 5894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453927580178965e-05,
+      "loss": 0.5332,
+      "step": 5895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453328901730122e-05,
+      "loss": 0.5707,
+      "step": 5896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452730117107e-05,
+      "loss": 0.4891,
+      "step": 5897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845213122631711e-05,
+      "loss": 0.5354,
+      "step": 5898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845153222936798e-05,
+      "loss": 0.6045,
+      "step": 5899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845093312626713e-05,
+      "loss": 0.5612,
+      "step": 5900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845033391702209e-05,
+      "loss": 0.5532,
+      "step": 5901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844973460164038e-05,
+      "loss": 0.5273,
+      "step": 5902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844913518012953e-05,
+      "loss": 0.5501,
+      "step": 5903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448535652497073e-05,
+      "loss": 0.504,
+      "step": 5904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447936018750533e-05,
+      "loss": 0.5631,
+      "step": 5905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844733627889744e-05,
+      "loss": 0.5551,
+      "step": 5906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446736432945335e-05,
+      "loss": 0.5706,
+      "step": 5907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446136480901744e-05,
+      "loss": 0.5742,
+      "step": 5908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844553642277421e-05,
+      "loss": 0.5369,
+      "step": 5909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844493625857026e-05,
+      "loss": 0.5549,
+      "step": 5910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844433598829744e-05,
+      "loss": 0.5996,
+      "step": 5911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443735611963286e-05,
+      "loss": 0.5589,
+      "step": 5912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443135129575345e-05,
+      "loss": 0.6384,
+      "step": 5913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844253454114115e-05,
+      "loss": 0.6365,
+      "step": 5914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441933846668246e-05,
+      "loss": 0.5509,
+      "step": 5915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441333046164183e-05,
+      "loss": 0.5942,
+      "step": 5916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8440732139636504e-05,
+      "loss": 0.586,
+      "step": 5917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844013112709276e-05,
+      "loss": 0.5335,
+      "step": 5918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439530008540488e-05,
+      "loss": 0.5365,
+      "step": 5919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438928783987252e-05,
+      "loss": 0.6026,
+      "step": 5920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438327453440596e-05,
+      "loss": 0.6448,
+      "step": 5921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437726016908074e-05,
+      "loss": 0.494,
+      "step": 5922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437124474397247e-05,
+      "loss": 0.587,
+      "step": 5923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8436522825915656e-05,
+      "loss": 0.6116,
+      "step": 5924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843592107147087e-05,
+      "loss": 0.6223,
+      "step": 5925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843531921107044e-05,
+      "loss": 0.5095,
+      "step": 5926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843471724472193e-05,
+      "loss": 0.5542,
+      "step": 5927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84341151724329e-05,
+      "loss": 0.4753,
+      "step": 5928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843351299421091e-05,
+      "loss": 0.5419,
+      "step": 5929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432910710063523e-05,
+      "loss": 0.5984,
+      "step": 5930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843230831999831e-05,
+      "loss": 0.5752,
+      "step": 5931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843170582402283e-05,
+      "loss": 0.5691,
+      "step": 5932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431103222144648e-05,
+      "loss": 0.5376,
+      "step": 5933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843050051437134e-05,
+      "loss": 0.5555,
+      "step": 5934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842989770071048e-05,
+      "loss": 0.5981,
+      "step": 5935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429294781169624e-05,
+      "loss": 0.581,
+      "step": 5936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428691755756356e-05,
+      "loss": 0.5892,
+      "step": 5937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842808862447825e-05,
+      "loss": 0.6161,
+      "step": 5938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427485387342873e-05,
+      "loss": 0.5569,
+      "step": 5939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426882044357813e-05,
+      "loss": 0.5415,
+      "step": 5940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426278595530644e-05,
+      "loss": 0.6048,
+      "step": 5941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842567504086894e-05,
+      "loss": 0.5427,
+      "step": 5942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425071380380286e-05,
+      "loss": 0.6072,
+      "step": 5943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8424467614072265e-05,
+      "loss": 0.6147,
+      "step": 5944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423863741952456e-05,
+      "loss": 0.5507,
+      "step": 5945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842325976402845e-05,
+      "loss": 0.5308,
+      "step": 5946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842265568030783e-05,
+      "loss": 0.5519,
+      "step": 5947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422051490798178e-05,
+      "loss": 0.5602,
+      "step": 5948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842144719550709e-05,
+      "loss": 0.6188,
+      "step": 5949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842084279444216e-05,
+      "loss": 0.5821,
+      "step": 5950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420238287610966e-05,
+      "loss": 0.5773,
+      "step": 5951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419633675021107e-05,
+      "loss": 0.5522,
+      "step": 5952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419028956680177e-05,
+      "loss": 0.5914,
+      "step": 5953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418424132595776e-05,
+      "loss": 0.5574,
+      "step": 5954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417819202775497e-05,
+      "loss": 0.5506,
+      "step": 5955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417214167226932e-05,
+      "loss": 0.6064,
+      "step": 5956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841660902595769e-05,
+      "loss": 0.5236,
+      "step": 5957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416003778975367e-05,
+      "loss": 0.5654,
+      "step": 5958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415398426287563e-05,
+      "loss": 0.5221,
+      "step": 5959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414792967901882e-05,
+      "loss": 0.5905,
+      "step": 5960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414187403825934e-05,
+      "loss": 0.6794,
+      "step": 5961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413581734067318e-05,
+      "loss": 0.5456,
+      "step": 5962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412975958633648e-05,
+      "loss": 0.5105,
+      "step": 5963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412370077532525e-05,
+      "loss": 0.5708,
+      "step": 5964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411764090771564e-05,
+      "loss": 0.592,
+      "step": 5965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411157998358375e-05,
+      "loss": 0.5323,
+      "step": 5966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841055180030057e-05,
+      "loss": 0.57,
+      "step": 5967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409945496605765e-05,
+      "loss": 0.5632,
+      "step": 5968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840933908728157e-05,
+      "loss": 0.5551,
+      "step": 5969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408732572335612e-05,
+      "loss": 0.576,
+      "step": 5970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408125951775496e-05,
+      "loss": 0.6334,
+      "step": 5971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407519225608847e-05,
+      "loss": 0.5936,
+      "step": 5972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406912393843288e-05,
+      "loss": 0.5403,
+      "step": 5973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840630545648644e-05,
+      "loss": 0.5539,
+      "step": 5974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840569841354592e-05,
+      "loss": 0.569,
+      "step": 5975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840509126502936e-05,
+      "loss": 0.5788,
+      "step": 5976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840448401094438e-05,
+      "loss": 0.648,
+      "step": 5977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403876651298615e-05,
+      "loss": 0.5499,
+      "step": 5978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403269186099683e-05,
+      "loss": 0.5534,
+      "step": 5979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402661615355225e-05,
+      "loss": 0.5778,
+      "step": 5980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402053939072865e-05,
+      "loss": 0.5248,
+      "step": 5981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401446157260237e-05,
+      "loss": 0.6278,
+      "step": 5982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840083826992497e-05,
+      "loss": 0.5695,
+      "step": 5983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400230277074712e-05,
+      "loss": 0.6179,
+      "step": 5984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399622178717083e-05,
+      "loss": 0.5614,
+      "step": 5985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399013974859733e-05,
+      "loss": 0.5325,
+      "step": 5986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83984056655103e-05,
+      "loss": 0.6197,
+      "step": 5987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397797250676416e-05,
+      "loss": 0.4748,
+      "step": 5988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839718873036573e-05,
+      "loss": 0.5695,
+      "step": 5989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396580104585883e-05,
+      "loss": 0.5849,
+      "step": 5990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395971373344522e-05,
+      "loss": 0.5897,
+      "step": 5991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839536253664929e-05,
+      "loss": 0.6094,
+      "step": 5992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839475359450783e-05,
+      "loss": 0.5459,
+      "step": 5993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394144546927803e-05,
+      "loss": 0.5973,
+      "step": 5994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393535393916845e-05,
+      "loss": 0.5232,
+      "step": 5995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839292613548261e-05,
+      "loss": 0.5193,
+      "step": 5996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392316771632758e-05,
+      "loss": 0.59,
+      "step": 5997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839170730237493e-05,
+      "loss": 0.6644,
+      "step": 5998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391097727716796e-05,
+      "loss": 0.6047,
+      "step": 5999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390488047666e-05,
+      "loss": 0.6035,
+      "step": 6000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389878262230207e-05,
+      "loss": 0.5845,
+      "step": 6001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389268371417072e-05,
+      "loss": 0.5983,
+      "step": 6002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388658375234255e-05,
+      "loss": 0.5399,
+      "step": 6003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388048273689422e-05,
+      "loss": 0.5122,
+      "step": 6004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838743806679023e-05,
+      "loss": 0.6013,
+      "step": 6005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386827754544348e-05,
+      "loss": 0.5486,
+      "step": 6006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386217336959437e-05,
+      "loss": 0.543,
+      "step": 6007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385606814043167e-05,
+      "loss": 0.537,
+      "step": 6008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838499618580321e-05,
+      "loss": 0.5601,
+      "step": 6009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838438545224723e-05,
+      "loss": 0.5266,
+      "step": 6010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383774613382894e-05,
+      "loss": 0.5633,
+      "step": 6011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383163669217885e-05,
+      "loss": 0.5075,
+      "step": 6012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8382552619759867e-05,
+      "loss": 0.6175,
+      "step": 6013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838194146501652e-05,
+      "loss": 0.5526,
+      "step": 6014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838133020499552e-05,
+      "loss": 0.5483,
+      "step": 6015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380718839704545e-05,
+      "loss": 0.6162,
+      "step": 6016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838010736915127e-05,
+      "loss": 0.5593,
+      "step": 6017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379495793343376e-05,
+      "loss": 0.6142,
+      "step": 6018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378884112288543e-05,
+      "loss": 0.5671,
+      "step": 6019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837827232599446e-05,
+      "loss": 0.5975,
+      "step": 6020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837766043446881e-05,
+      "loss": 0.563,
+      "step": 6021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377048437719273e-05,
+      "loss": 0.5761,
+      "step": 6022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376436335753534e-05,
+      "loss": 0.601,
+      "step": 6023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375824128579293e-05,
+      "loss": 0.5587,
+      "step": 6024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375211816204224e-05,
+      "loss": 0.6001,
+      "step": 6025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374599398636032e-05,
+      "loss": 0.5813,
+      "step": 6026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83739868758824e-05,
+      "loss": 0.5168,
+      "step": 6027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373374247951024e-05,
+      "loss": 0.492,
+      "step": 6028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372761514849598e-05,
+      "loss": 0.597,
+      "step": 6029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837214867658582e-05,
+      "loss": 0.6057,
+      "step": 6030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371535733167387e-05,
+      "loss": 0.5592,
+      "step": 6031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370922684601993e-05,
+      "loss": 0.508,
+      "step": 6032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837030953089734e-05,
+      "loss": 0.6312,
+      "step": 6033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369696272061132e-05,
+      "loss": 0.5473,
+      "step": 6034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369082908101072e-05,
+      "loss": 0.5376,
+      "step": 6035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836846943902486e-05,
+      "loss": 0.5912,
+      "step": 6036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367855864840207e-05,
+      "loss": 0.5645,
+      "step": 6037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367242185554813e-05,
+      "loss": 0.5508,
+      "step": 6038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366628401176387e-05,
+      "loss": 0.5818,
+      "step": 6039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836601451171264e-05,
+      "loss": 0.5261,
+      "step": 6040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365400517171282e-05,
+      "loss": 0.5534,
+      "step": 6041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836478641756003e-05,
+      "loss": 0.6284,
+      "step": 6042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364172212886586e-05,
+      "loss": 0.5207,
+      "step": 6043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363557903158672e-05,
+      "loss": 0.5878,
+      "step": 6044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362943488384003e-05,
+      "loss": 0.5184,
+      "step": 6045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362328968570295e-05,
+      "loss": 0.6184,
+      "step": 6046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361714343725268e-05,
+      "loss": 0.5074,
+      "step": 6047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361099613856642e-05,
+      "loss": 0.577,
+      "step": 6048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360484778972133e-05,
+      "loss": 0.5454,
+      "step": 6049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835986983907947e-05,
+      "loss": 0.5984,
+      "step": 6050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359254794186368e-05,
+      "loss": 0.5582,
+      "step": 6051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358639644300566e-05,
+      "loss": 0.5855,
+      "step": 6052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358024389429772e-05,
+      "loss": 0.5559,
+      "step": 6053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835740902958173e-05,
+      "loss": 0.7109,
+      "step": 6054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356793564764165e-05,
+      "loss": 0.5502,
+      "step": 6055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83561779949848e-05,
+      "loss": 0.5131,
+      "step": 6056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355562320251374e-05,
+      "loss": 0.5586,
+      "step": 6057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354946540571618e-05,
+      "loss": 0.5386,
+      "step": 6058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835433065595326e-05,
+      "loss": 0.557,
+      "step": 6059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353714666404047e-05,
+      "loss": 0.5636,
+      "step": 6060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353098571931706e-05,
+      "loss": 0.5456,
+      "step": 6061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835248237254398e-05,
+      "loss": 0.5588,
+      "step": 6062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835186606824861e-05,
+      "loss": 0.6495,
+      "step": 6063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351249659053333e-05,
+      "loss": 0.5864,
+      "step": 6064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835063314496589e-05,
+      "loss": 0.5152,
+      "step": 6065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350016525994032e-05,
+      "loss": 0.6513,
+      "step": 6066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349399802145495e-05,
+      "loss": 0.6049,
+      "step": 6067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834878297342803e-05,
+      "loss": 0.5431,
+      "step": 6068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348166039849382e-05,
+      "loss": 0.6407,
+      "step": 6069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347549001417304e-05,
+      "loss": 0.59,
+      "step": 6070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834693185813954e-05,
+      "loss": 0.4807,
+      "step": 6071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346314610023848e-05,
+      "loss": 0.5325,
+      "step": 6072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8345697257077974e-05,
+      "loss": 0.606,
+      "step": 6073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8345079799309675e-05,
+      "loss": 0.5582,
+      "step": 6074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344462236726708e-05,
+      "loss": 0.5969,
+      "step": 6075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343844569336827e-05,
+      "loss": 0.5929,
+      "step": 6076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834322679714779e-05,
+      "loss": 0.5439,
+      "step": 6077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834260892016736e-05,
+      "loss": 0.5295,
+      "step": 6078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834199093840329e-05,
+      "loss": 0.5556,
+      "step": 6079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341372851863353e-05,
+      "loss": 0.5168,
+      "step": 6080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340754660555303e-05,
+      "loss": 0.5876,
+      "step": 6081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340136364486906e-05,
+      "loss": 0.4952,
+      "step": 6082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339517963665928e-05,
+      "loss": 0.5988,
+      "step": 6083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833889945810014e-05,
+      "loss": 0.5879,
+      "step": 6084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338280847797305e-05,
+      "loss": 0.5686,
+      "step": 6085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83376621327652e-05,
+      "loss": 0.533,
+      "step": 6086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337043313011588e-05,
+      "loss": 0.5216,
+      "step": 6087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336424388544247e-05,
+      "loss": 0.5801,
+      "step": 6088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335805359370945e-05,
+      "loss": 0.6224,
+      "step": 6089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335186225499465e-05,
+      "loss": 0.534,
+      "step": 6090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334566986937578e-05,
+      "loss": 0.5223,
+      "step": 6091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333947643693065e-05,
+      "loss": 0.5297,
+      "step": 6092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333328195773698e-05,
+      "loss": 0.5592,
+      "step": 6093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332708643187265e-05,
+      "loss": 0.5951,
+      "step": 6094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332088985941544e-05,
+      "loss": 0.6035,
+      "step": 6095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331469224044322e-05,
+      "loss": 0.5654,
+      "step": 6096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330849357503376e-05,
+      "loss": 0.5726,
+      "step": 6097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330229386326498e-05,
+      "loss": 0.5548,
+      "step": 6098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832960931052147e-05,
+      "loss": 0.5739,
+      "step": 6099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328989130096082e-05,
+      "loss": 0.654,
+      "step": 6100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328368845058125e-05,
+      "loss": 0.5547,
+      "step": 6101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832774845541539e-05,
+      "loss": 0.4961,
+      "step": 6102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327127961175668e-05,
+      "loss": 0.6032,
+      "step": 6103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8326507362346753e-05,
+      "loss": 0.6119,
+      "step": 6104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832588665893644e-05,
+      "loss": 0.5401,
+      "step": 6105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832526585095252e-05,
+      "loss": 0.5152,
+      "step": 6106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324644938402796e-05,
+      "loss": 0.5728,
+      "step": 6107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324023921295066e-05,
+      "loss": 0.6075,
+      "step": 6108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832340279963713e-05,
+      "loss": 0.5274,
+      "step": 6109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322781573436787e-05,
+      "loss": 0.5474,
+      "step": 6110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322160242701844e-05,
+      "loss": 0.583,
+      "step": 6111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83215388074401e-05,
+      "loss": 0.6051,
+      "step": 6112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320917267659365e-05,
+      "loss": 0.5312,
+      "step": 6113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832029562336744e-05,
+      "loss": 0.5131,
+      "step": 6114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831967387457214e-05,
+      "loss": 0.5878,
+      "step": 6115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319052021281267e-05,
+      "loss": 0.5507,
+      "step": 6116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318430063502635e-05,
+      "loss": 0.5869,
+      "step": 6117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831780800124406e-05,
+      "loss": 0.6292,
+      "step": 6118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317185834513343e-05,
+      "loss": 0.574,
+      "step": 6119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316563563318312e-05,
+      "loss": 0.5461,
+      "step": 6120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831594118766677e-05,
+      "loss": 0.5645,
+      "step": 6121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315318707566546e-05,
+      "loss": 0.6131,
+      "step": 6122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.6157,
+      "step": 6123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831407343405131e-05,
+      "loss": 0.6232,
+      "step": 6124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313450640651943e-05,
+      "loss": 0.6267,
+      "step": 6125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312827742835168e-05,
+      "loss": 0.5677,
+      "step": 6126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831220474060881e-05,
+      "loss": 0.5246,
+      "step": 6127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311581633980692e-05,
+      "loss": 0.5432,
+      "step": 6128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310958422958648e-05,
+      "loss": 0.517,
+      "step": 6129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83103351075505e-05,
+      "loss": 0.6081,
+      "step": 6130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309711687764077e-05,
+      "loss": 0.5468,
+      "step": 6131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309088163607207e-05,
+      "loss": 0.564,
+      "step": 6132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308464535087725e-05,
+      "loss": 0.5507,
+      "step": 6133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307840802213467e-05,
+      "loss": 0.6361,
+      "step": 6134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830721696499226e-05,
+      "loss": 0.5772,
+      "step": 6135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8306593023431946e-05,
+      "loss": 0.5697,
+      "step": 6136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305968977540355e-05,
+      "loss": 0.6302,
+      "step": 6137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830534482732533e-05,
+      "loss": 0.5399,
+      "step": 6138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830472057279471e-05,
+      "loss": 0.5685,
+      "step": 6139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830409621395633e-05,
+      "loss": 0.576,
+      "step": 6140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830347175081804e-05,
+      "loss": 0.5785,
+      "step": 6141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830284718338768e-05,
+      "loss": 0.5129,
+      "step": 6142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8302222511673094e-05,
+      "loss": 0.6136,
+      "step": 6143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301597735682128e-05,
+      "loss": 0.5648,
+      "step": 6144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830097285542263e-05,
+      "loss": 0.624,
+      "step": 6145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300347870902448e-05,
+      "loss": 0.5886,
+      "step": 6146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299722782129428e-05,
+      "loss": 0.5899,
+      "step": 6147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829909758911143e-05,
+      "loss": 0.5474,
+      "step": 6148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82984722918563e-05,
+      "loss": 0.5979,
+      "step": 6149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829784689037189e-05,
+      "loss": 0.4981,
+      "step": 6150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297221384666063e-05,
+      "loss": 0.5451,
+      "step": 6151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8296595774746668e-05,
+      "loss": 0.5526,
+      "step": 6152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829597006062157e-05,
+      "loss": 0.5759,
+      "step": 6153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295344242298617e-05,
+      "loss": 0.5599,
+      "step": 6154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294718319785677e-05,
+      "loss": 0.502,
+      "step": 6155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294092293090615e-05,
+      "loss": 0.5505,
+      "step": 6156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293466162221284e-05,
+      "loss": 0.5253,
+      "step": 6157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292839927185556e-05,
+      "loss": 0.5997,
+      "step": 6158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292213587991295e-05,
+      "loss": 0.5579,
+      "step": 6159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8291587144646367e-05,
+      "loss": 0.5628,
+      "step": 6160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290960597158637e-05,
+      "loss": 0.6182,
+      "step": 6161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290333945535978e-05,
+      "loss": 0.5378,
+      "step": 6162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289707189786264e-05,
+      "loss": 0.5415,
+      "step": 6163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828908032991736e-05,
+      "loss": 0.5868,
+      "step": 6164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8288453365937143e-05,
+      "loss": 0.5568,
+      "step": 6165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287826297853487e-05,
+      "loss": 0.5948,
+      "step": 6166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287199125674272e-05,
+      "loss": 0.6445,
+      "step": 6167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828657184940737e-05,
+      "loss": 0.5499,
+      "step": 6168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828594446906066e-05,
+      "loss": 0.5723,
+      "step": 6169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8285316984642025e-05,
+      "loss": 0.5314,
+      "step": 6170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284689396159347e-05,
+      "loss": 0.587,
+      "step": 6171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82840617036205e-05,
+      "loss": 0.5294,
+      "step": 6172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828343390703338e-05,
+      "loss": 0.6047,
+      "step": 6173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282806006405863e-05,
+      "loss": 0.5861,
+      "step": 6174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828217800174584e-05,
+      "loss": 0.5717,
+      "step": 6175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281549893061194e-05,
+      "loss": 0.6107,
+      "step": 6176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280921680359822e-05,
+      "loss": 0.466,
+      "step": 6177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280293363649607e-05,
+      "loss": 0.6037,
+      "step": 6178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279664942938448e-05,
+      "loss": 0.5551,
+      "step": 6179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827903641823423e-05,
+      "loss": 0.5469,
+      "step": 6180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827840778954485e-05,
+      "loss": 0.6047,
+      "step": 6181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277779056878207e-05,
+      "loss": 0.4952,
+      "step": 6182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277150220242195e-05,
+      "loss": 0.5244,
+      "step": 6183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827652127964471e-05,
+      "loss": 0.5834,
+      "step": 6184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275892235093656e-05,
+      "loss": 0.5472,
+      "step": 6185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827526308659693e-05,
+      "loss": 0.638,
+      "step": 6186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827463383416244e-05,
+      "loss": 0.6393,
+      "step": 6187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274004477798085e-05,
+      "loss": 0.5604,
+      "step": 6188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827337501751177e-05,
+      "loss": 0.5648,
+      "step": 6189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82727454533114e-05,
+      "loss": 0.52,
+      "step": 6190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272115785204884e-05,
+      "loss": 0.4844,
+      "step": 6191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271486013200132e-05,
+      "loss": 0.5855,
+      "step": 6192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827085613730505e-05,
+      "loss": 0.6006,
+      "step": 6193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270226157527557e-05,
+      "loss": 0.5764,
+      "step": 6194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269596073875555e-05,
+      "loss": 0.5148,
+      "step": 6195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268965886356965e-05,
+      "loss": 0.624,
+      "step": 6196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268335594979698e-05,
+      "loss": 0.6134,
+      "step": 6197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267705199751675e-05,
+      "loss": 0.5344,
+      "step": 6198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267074700680815e-05,
+      "loss": 0.5762,
+      "step": 6199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266444097775028e-05,
+      "loss": 0.514,
+      "step": 6200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826581339104224e-05,
+      "loss": 0.5728,
+      "step": 6201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8265182580490374e-05,
+      "loss": 0.6393,
+      "step": 6202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264551666127354e-05,
+      "loss": 0.6265,
+      "step": 6203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8263920647961104e-05,
+      "loss": 0.548,
+      "step": 6204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8263289525999544e-05,
+      "loss": 0.5743,
+      "step": 6205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262658300250607e-05,
+      "loss": 0.534,
+      "step": 6206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262026970722218e-05,
+      "loss": 0.5416,
+      "step": 6207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261395537422307e-05,
+      "loss": 0.5775,
+      "step": 6208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260764000358804e-05,
+      "loss": 0.6004,
+      "step": 6209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260132359539647e-05,
+      "loss": 0.5788,
+      "step": 6210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825950061497276e-05,
+      "loss": 0.5942,
+      "step": 6211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258868766666083e-05,
+      "loss": 0.5557,
+      "step": 6212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258236814627553e-05,
+      "loss": 0.5762,
+      "step": 6213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257604758865112e-05,
+      "loss": 0.4964,
+      "step": 6214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8256972599386685e-05,
+      "loss": 0.5878,
+      "step": 6215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825634033620022e-05,
+      "loss": 0.5923,
+      "step": 6216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255707969313664e-05,
+      "loss": 0.6198,
+      "step": 6217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825507549873495e-05,
+      "loss": 0.5393,
+      "step": 6218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254442924472024e-05,
+      "loss": 0.6056,
+      "step": 6219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253810246532834e-05,
+      "loss": 0.5083,
+      "step": 6220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253177464925323e-05,
+      "loss": 0.604,
+      "step": 6221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252544579657442e-05,
+      "loss": 0.4871,
+      "step": 6222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251911590737143e-05,
+      "loss": 0.5789,
+      "step": 6223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251278498172364e-05,
+      "loss": 0.6239,
+      "step": 6224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8250645301971072e-05,
+      "loss": 0.5855,
+      "step": 6225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825001200214121e-05,
+      "loss": 0.6466,
+      "step": 6226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249378598690736e-05,
+      "loss": 0.5716,
+      "step": 6227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248745091627603e-05,
+      "loss": 0.5561,
+      "step": 6228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248111480959774e-05,
+      "loss": 0.6071,
+      "step": 6229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247477766695197e-05,
+      "loss": 0.5593,
+      "step": 6230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246843948841842e-05,
+      "loss": 0.5389,
+      "step": 6231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246210027407664e-05,
+      "loss": 0.5837,
+      "step": 6232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245576002400628e-05,
+      "loss": 0.5509,
+      "step": 6233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824494187382869e-05,
+      "loss": 0.6017,
+      "step": 6234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244307641699825e-05,
+      "loss": 0.5737,
+      "step": 6235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824367330602199e-05,
+      "loss": 0.5254,
+      "step": 6236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824303886680316e-05,
+      "loss": 0.5438,
+      "step": 6237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242404324051303e-05,
+      "loss": 0.6509,
+      "step": 6238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824176967777438e-05,
+      "loss": 0.5959,
+      "step": 6239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8241134927980373e-05,
+      "loss": 0.5883,
+      "step": 6240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240500074677246e-05,
+      "loss": 0.5288,
+      "step": 6241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823986511787298e-05,
+      "loss": 0.5764,
+      "step": 6242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239230057575542e-05,
+      "loss": 0.5804,
+      "step": 6243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823859489379292e-05,
+      "loss": 0.5778,
+      "step": 6244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237959626533078e-05,
+      "loss": 0.6005,
+      "step": 6245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237324255804006e-05,
+      "loss": 0.6153,
+      "step": 6246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236688781613677e-05,
+      "loss": 0.6113,
+      "step": 6247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823605320397008e-05,
+      "loss": 0.5731,
+      "step": 6248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823541752288119e-05,
+      "loss": 0.5722,
+      "step": 6249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234781738354993e-05,
+      "loss": 0.4911,
+      "step": 6250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234145850399478e-05,
+      "loss": 0.4647,
+      "step": 6251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823350985902263e-05,
+      "loss": 0.5288,
+      "step": 6252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232873764232437e-05,
+      "loss": 0.6227,
+      "step": 6253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823223756603689e-05,
+      "loss": 0.5405,
+      "step": 6254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231601264443974e-05,
+      "loss": 0.587,
+      "step": 6255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823096485946169e-05,
+      "loss": 0.5833,
+      "step": 6256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230328351098024e-05,
+      "loss": 0.6171,
+      "step": 6257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229691739360973e-05,
+      "loss": 0.6217,
+      "step": 6258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229055024258534e-05,
+      "loss": 0.5378,
+      "step": 6259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228418205798705e-05,
+      "loss": 0.6195,
+      "step": 6260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227781283989476e-05,
+      "loss": 0.5567,
+      "step": 6261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227144258838856e-05,
+      "loss": 0.5534,
+      "step": 6262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226507130354845e-05,
+      "loss": 0.5123,
+      "step": 6263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225869898545445e-05,
+      "loss": 0.5496,
+      "step": 6264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225232563418656e-05,
+      "loss": 0.5756,
+      "step": 6265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8224595124982484e-05,
+      "loss": 0.6317,
+      "step": 6266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822395758324494e-05,
+      "loss": 0.5646,
+      "step": 6267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223319938214022e-05,
+      "loss": 0.5393,
+      "step": 6268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822268218989775e-05,
+      "loss": 0.55,
+      "step": 6269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222044338304132e-05,
+      "loss": 0.5321,
+      "step": 6270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221406383441174e-05,
+      "loss": 0.5464,
+      "step": 6271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220768325316887e-05,
+      "loss": 0.5432,
+      "step": 6272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220130163939297e-05,
+      "loss": 0.5755,
+      "step": 6273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219491899316408e-05,
+      "loss": 0.5465,
+      "step": 6274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821885353145624e-05,
+      "loss": 0.5865,
+      "step": 6275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821821506036681e-05,
+      "loss": 0.5337,
+      "step": 6276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217576486056143e-05,
+      "loss": 0.5741,
+      "step": 6277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821693780853225e-05,
+      "loss": 0.5869,
+      "step": 6278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216299027803156e-05,
+      "loss": 0.6003,
+      "step": 6279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821566014387689e-05,
+      "loss": 0.6058,
+      "step": 6280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215021156761477e-05,
+      "loss": 0.5288,
+      "step": 6281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821438206646493e-05,
+      "loss": 0.5669,
+      "step": 6282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821374287299529e-05,
+      "loss": 0.5673,
+      "step": 6283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821310357636058e-05,
+      "loss": 0.5302,
+      "step": 6284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212464176568826e-05,
+      "loss": 0.5754,
+      "step": 6285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211824673628064e-05,
+      "loss": 0.5262,
+      "step": 6286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821118506754632e-05,
+      "loss": 0.5302,
+      "step": 6287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210545358331635e-05,
+      "loss": 0.5687,
+      "step": 6288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820990554599204e-05,
+      "loss": 0.6032,
+      "step": 6289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820926563053557e-05,
+      "loss": 0.5739,
+      "step": 6290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208625611970268e-05,
+      "loss": 0.5895,
+      "step": 6291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207985490304162e-05,
+      "loss": 0.5718,
+      "step": 6292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207345265545304e-05,
+      "loss": 0.5726,
+      "step": 6293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820670493770173e-05,
+      "loss": 0.5513,
+      "step": 6294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820606450678148e-05,
+      "loss": 0.5368,
+      "step": 6295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82054239727926e-05,
+      "loss": 0.5581,
+      "step": 6296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204783335743138e-05,
+      "loss": 0.6258,
+      "step": 6297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204142595641133e-05,
+      "loss": 0.5812,
+      "step": 6298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203501752494643e-05,
+      "loss": 0.6438,
+      "step": 6299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202860806311708e-05,
+      "loss": 0.5382,
+      "step": 6300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202219757100385e-05,
+      "loss": 0.4966,
+      "step": 6301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201578604868716e-05,
+      "loss": 0.5197,
+      "step": 6302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200937349624767e-05,
+      "loss": 0.5839,
+      "step": 6303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200295991376583e-05,
+      "loss": 0.4855,
+      "step": 6304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819965453013222e-05,
+      "loss": 0.585,
+      "step": 6305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199012965899743e-05,
+      "loss": 0.5081,
+      "step": 6306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81983712986872e-05,
+      "loss": 0.503,
+      "step": 6307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197729528502653e-05,
+      "loss": 0.5964,
+      "step": 6308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197087655354165e-05,
+      "loss": 0.5885,
+      "step": 6309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81964456792498e-05,
+      "loss": 0.6128,
+      "step": 6310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195803600197613e-05,
+      "loss": 0.6084,
+      "step": 6311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195161418205677e-05,
+      "loss": 0.5644,
+      "step": 6312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8194519133282056e-05,
+      "loss": 0.621,
+      "step": 6313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193876745434814e-05,
+      "loss": 0.5173,
+      "step": 6314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819323425467202e-05,
+      "loss": 0.5471,
+      "step": 6315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192591661001745e-05,
+      "loss": 0.5003,
+      "step": 6316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191948964432062e-05,
+      "loss": 0.5462,
+      "step": 6317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819130616497104e-05,
+      "loss": 0.5829,
+      "step": 6318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190663262626754e-05,
+      "loss": 0.5881,
+      "step": 6319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190020257407277e-05,
+      "loss": 0.6547,
+      "step": 6320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818937714932069e-05,
+      "loss": 0.6326,
+      "step": 6321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188733938375064e-05,
+      "loss": 0.6029,
+      "step": 6322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188090624578482e-05,
+      "loss": 0.6049,
+      "step": 6323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187447207939026e-05,
+      "loss": 0.5446,
+      "step": 6324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186803688464773e-05,
+      "loss": 0.4419,
+      "step": 6325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186160066163806e-05,
+      "loss": 0.5701,
+      "step": 6326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818551634104421e-05,
+      "loss": 0.553,
+      "step": 6327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818487251311407e-05,
+      "loss": 0.5693,
+      "step": 6328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184228582381475e-05,
+      "loss": 0.5812,
+      "step": 6329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818358454885451e-05,
+      "loss": 0.5023,
+      "step": 6330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182940412541262e-05,
+      "loss": 0.578,
+      "step": 6331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818229617344983e-05,
+      "loss": 0.5725,
+      "step": 6332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181651831588296e-05,
+      "loss": 0.6321,
+      "step": 6333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181007386964756e-05,
+      "loss": 0.5396,
+      "step": 6334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8180362839587306e-05,
+      "loss": 0.5723,
+      "step": 6335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179718189464042e-05,
+      "loss": 0.5913,
+      "step": 6336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179073436603058e-05,
+      "loss": 0.4873,
+      "step": 6337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178428581012454e-05,
+      "loss": 0.5643,
+      "step": 6338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177783622700328e-05,
+      "loss": 0.5406,
+      "step": 6339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177138561674784e-05,
+      "loss": 0.5534,
+      "step": 6340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8176493397943917e-05,
+      "loss": 0.5703,
+      "step": 6341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817584813151584e-05,
+      "loss": 0.557,
+      "step": 6342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817520276239865e-05,
+      "loss": 0.4872,
+      "step": 6343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174557290600456e-05,
+      "loss": 0.6608,
+      "step": 6344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817391171612936e-05,
+      "loss": 0.5492,
+      "step": 6345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173266038993485e-05,
+      "loss": 0.5704,
+      "step": 6346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8172620259200922e-05,
+      "loss": 0.575,
+      "step": 6347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171974376759795e-05,
+      "loss": 0.5039,
+      "step": 6348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171328391678213e-05,
+      "loss": 0.574,
+      "step": 6349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170682303964288e-05,
+      "loss": 0.5375,
+      "step": 6350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817003611362613e-05,
+      "loss": 0.5915,
+      "step": 6351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169389820671868e-05,
+      "loss": 0.5997,
+      "step": 6352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816874342510961e-05,
+      "loss": 0.5696,
+      "step": 6353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816809692694748e-05,
+      "loss": 0.5456,
+      "step": 6354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816745032619359e-05,
+      "loss": 0.565,
+      "step": 6355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166803622856067e-05,
+      "loss": 0.5416,
+      "step": 6356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166156816943036e-05,
+      "loss": 0.5737,
+      "step": 6357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165509908462618e-05,
+      "loss": 0.5351,
+      "step": 6358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164862897422936e-05,
+      "loss": 0.5373,
+      "step": 6359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164215783832118e-05,
+      "loss": 0.5965,
+      "step": 6360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81635685676983e-05,
+      "loss": 0.5547,
+      "step": 6361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162921249029593e-05,
+      "loss": 0.6983,
+      "step": 6362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162273827834146e-05,
+      "loss": 0.5547,
+      "step": 6363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816162630412008e-05,
+      "loss": 0.5617,
+      "step": 6364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160978677895525e-05,
+      "loss": 0.5153,
+      "step": 6365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816033094916863e-05,
+      "loss": 0.5733,
+      "step": 6366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159683117947514e-05,
+      "loss": 0.5289,
+      "step": 6367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159035184240324e-05,
+      "loss": 0.6194,
+      "step": 6368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158387148055196e-05,
+      "loss": 0.617,
+      "step": 6369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157739009400267e-05,
+      "loss": 0.5434,
+      "step": 6370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815709076828368e-05,
+      "loss": 0.5774,
+      "step": 6371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156442424713575e-05,
+      "loss": 0.5752,
+      "step": 6372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155793978698097e-05,
+      "loss": 0.6377,
+      "step": 6373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815514543024539e-05,
+      "loss": 0.5441,
+      "step": 6374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8154496779363603e-05,
+      "loss": 0.5239,
+      "step": 6375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153848026060873e-05,
+      "loss": 0.51,
+      "step": 6376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153199170345358e-05,
+      "loss": 0.597,
+      "step": 6377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8152550212225206e-05,
+      "loss": 0.5579,
+      "step": 6378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151901151708565e-05,
+      "loss": 0.5972,
+      "step": 6379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151251988803592e-05,
+      "loss": 0.5944,
+      "step": 6380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150602723518434e-05,
+      "loss": 0.5132,
+      "step": 6381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814995335586125e-05,
+      "loss": 0.5417,
+      "step": 6382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149303885840196e-05,
+      "loss": 0.562,
+      "step": 6383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814865431346343e-05,
+      "loss": 0.5793,
+      "step": 6384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814800463873911e-05,
+      "loss": 0.6103,
+      "step": 6385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814735486167539e-05,
+      "loss": 0.6076,
+      "step": 6386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146704982280444e-05,
+      "loss": 0.5189,
+      "step": 6387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146055000562422e-05,
+      "loss": 0.5495,
+      "step": 6388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145404916529495e-05,
+      "loss": 0.5245,
+      "step": 6389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144754730189828e-05,
+      "loss": 0.564,
+      "step": 6390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144104441551583e-05,
+      "loss": 0.5993,
+      "step": 6391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814345405062293e-05,
+      "loss": 0.4979,
+      "step": 6392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142803557412037e-05,
+      "loss": 0.5076,
+      "step": 6393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142152961927077e-05,
+      "loss": 0.5511,
+      "step": 6394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141502264176218e-05,
+      "loss": 0.5522,
+      "step": 6395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814085146416764e-05,
+      "loss": 0.6051,
+      "step": 6396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8140200561909504e-05,
+      "loss": 0.6138,
+      "step": 6397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139549557409996e-05,
+      "loss": 0.6167,
+      "step": 6398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138898450677288e-05,
+      "loss": 0.6644,
+      "step": 6399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138247241719565e-05,
+      "loss": 0.5893,
+      "step": 6400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137595930545e-05,
+      "loss": 0.6139,
+      "step": 6401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813694451716177e-05,
+      "loss": 0.5633,
+      "step": 6402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813629300157806e-05,
+      "loss": 0.5821,
+      "step": 6403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135641383802062e-05,
+      "loss": 0.6085,
+      "step": 6404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8134989663841945e-05,
+      "loss": 0.5718,
+      "step": 6405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8134337841705908e-05,
+      "loss": 0.6578,
+      "step": 6406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133685917402125e-05,
+      "loss": 0.5804,
+      "step": 6407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81330338909388e-05,
+      "loss": 0.5952,
+      "step": 6408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132381762324107e-05,
+      "loss": 0.5804,
+      "step": 6409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131729531566247e-05,
+      "loss": 0.5753,
+      "step": 6410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131077198673408e-05,
+      "loss": 0.5765,
+      "step": 6411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130424763653786e-05,
+      "loss": 0.5541,
+      "step": 6412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812977222651557e-05,
+      "loss": 0.532,
+      "step": 6413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812911958726696e-05,
+      "loss": 0.6114,
+      "step": 6414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128466845916156e-05,
+      "loss": 0.5626,
+      "step": 6415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812781400247135e-05,
+      "loss": 0.594,
+      "step": 6416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127161056940744e-05,
+      "loss": 0.5308,
+      "step": 6417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8126508009332543e-05,
+      "loss": 0.563,
+      "step": 6418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125854859654944e-05,
+      "loss": 0.6274,
+      "step": 6419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125201607916153e-05,
+      "loss": 0.6062,
+      "step": 6420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124548254124376e-05,
+      "loss": 0.5264,
+      "step": 6421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812389479828782e-05,
+      "loss": 0.4959,
+      "step": 6422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123241240414688e-05,
+      "loss": 0.5226,
+      "step": 6423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122587580513186e-05,
+      "loss": 0.5079,
+      "step": 6424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121933818591536e-05,
+      "loss": 0.5916,
+      "step": 6425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812127995465794e-05,
+      "loss": 0.5271,
+      "step": 6426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120625988720616e-05,
+      "loss": 0.5753,
+      "step": 6427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811997192078777e-05,
+      "loss": 0.6258,
+      "step": 6428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119317750867624e-05,
+      "loss": 0.5837,
+      "step": 6429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118663478968392e-05,
+      "loss": 0.5929,
+      "step": 6430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118009105098293e-05,
+      "loss": 0.5802,
+      "step": 6431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811735462926554e-05,
+      "loss": 0.5803,
+      "step": 6432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116700051478365e-05,
+      "loss": 0.5738,
+      "step": 6433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811604537174498e-05,
+      "loss": 0.5995,
+      "step": 6434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115390590073612e-05,
+      "loss": 0.6055,
+      "step": 6435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114735706472482e-05,
+      "loss": 0.548,
+      "step": 6436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114080720949818e-05,
+      "loss": 0.5521,
+      "step": 6437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113425633513844e-05,
+      "loss": 0.4838,
+      "step": 6438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112770444172793e-05,
+      "loss": 0.5653,
+      "step": 6439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811211515293489e-05,
+      "loss": 0.5034,
+      "step": 6440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111459759808363e-05,
+      "loss": 0.5649,
+      "step": 6441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811080426480145e-05,
+      "loss": 0.5638,
+      "step": 6442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811014866792238e-05,
+      "loss": 0.5276,
+      "step": 6443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810949296917939e-05,
+      "loss": 0.5095,
+      "step": 6444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108837168580712e-05,
+      "loss": 0.509,
+      "step": 6445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108181266134586e-05,
+      "loss": 0.5877,
+      "step": 6446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810752526184925e-05,
+      "loss": 0.5758,
+      "step": 6447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810686915573294e-05,
+      "loss": 0.5835,
+      "step": 6448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106212947793902e-05,
+      "loss": 0.624,
+      "step": 6449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105556638040373e-05,
+      "loss": 0.5882,
+      "step": 6450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81049002264806e-05,
+      "loss": 0.5602,
+      "step": 6451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104243713122828e-05,
+      "loss": 0.4893,
+      "step": 6452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81035870979753e-05,
+      "loss": 0.5433,
+      "step": 6453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102930381046263e-05,
+      "loss": 0.5993,
+      "step": 6454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102273562343968e-05,
+      "loss": 0.5806,
+      "step": 6455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810161664187666e-05,
+      "loss": 0.5589,
+      "step": 6456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100959619652597e-05,
+      "loss": 0.5515,
+      "step": 6457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100302495680027e-05,
+      "loss": 0.6146,
+      "step": 6458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099645269967207e-05,
+      "loss": 0.6083,
+      "step": 6459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098987942522385e-05,
+      "loss": 0.5725,
+      "step": 6460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098330513353816e-05,
+      "loss": 0.5487,
+      "step": 6461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809767298246977e-05,
+      "loss": 0.5553,
+      "step": 6462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097015349878493e-05,
+      "loss": 0.4987,
+      "step": 6463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8096357615588253e-05,
+      "loss": 0.6883,
+      "step": 6464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095699779607306e-05,
+      "loss": 0.5084,
+      "step": 6465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809504184194392e-05,
+      "loss": 0.5437,
+      "step": 6466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809438380260635e-05,
+      "loss": 0.6079,
+      "step": 6467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093725661602867e-05,
+      "loss": 0.5773,
+      "step": 6468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093067418941738e-05,
+      "loss": 0.507,
+      "step": 6469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809240907463123e-05,
+      "loss": 0.5964,
+      "step": 6470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809175062867961e-05,
+      "loss": 0.5603,
+      "step": 6471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091092081095146e-05,
+      "loss": 0.5816,
+      "step": 6472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090433431886114e-05,
+      "loss": 0.5946,
+      "step": 6473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089774681060785e-05,
+      "loss": 0.5741,
+      "step": 6474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089115828627433e-05,
+      "loss": 0.5412,
+      "step": 6475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088456874594337e-05,
+      "loss": 0.5612,
+      "step": 6476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087797818969765e-05,
+      "loss": 0.5493,
+      "step": 6477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087138661762e-05,
+      "loss": 0.5608,
+      "step": 6478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086479402979323e-05,
+      "loss": 0.5988,
+      "step": 6479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808582004263001e-05,
+      "loss": 0.5755,
+      "step": 6480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085160580722344e-05,
+      "loss": 0.4375,
+      "step": 6481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084501017264607e-05,
+      "loss": 0.5661,
+      "step": 6482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083841352265085e-05,
+      "loss": 0.5414,
+      "step": 6483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083181585732066e-05,
+      "loss": 0.5664,
+      "step": 6484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808252171767383e-05,
+      "loss": 0.574,
+      "step": 6485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081861748098673e-05,
+      "loss": 0.5478,
+      "step": 6486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081201677014876e-05,
+      "loss": 0.5725,
+      "step": 6487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080541504430732e-05,
+      "loss": 0.5954,
+      "step": 6488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079881230354538e-05,
+      "loss": 0.5336,
+      "step": 6489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807922085479458e-05,
+      "loss": 0.4811,
+      "step": 6490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807856037775916e-05,
+      "loss": 0.5265,
+      "step": 6491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077899799256565e-05,
+      "loss": 0.5705,
+      "step": 6492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077239119295097e-05,
+      "loss": 0.4829,
+      "step": 6493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076578337883054e-05,
+      "loss": 0.5813,
+      "step": 6494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075917455028733e-05,
+      "loss": 0.4948,
+      "step": 6495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807525647074044e-05,
+      "loss": 0.5203,
+      "step": 6496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807459538502647e-05,
+      "loss": 0.5647,
+      "step": 6497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073934197895133e-05,
+      "loss": 0.5492,
+      "step": 6498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073272909354727e-05,
+      "loss": 0.5676,
+      "step": 6499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072611519413563e-05,
+      "loss": 0.5711,
+      "step": 6500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8071950028079944e-05,
+      "loss": 0.5076,
+      "step": 6501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8071288435362183e-05,
+      "loss": 0.5206,
+      "step": 6502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070626741268585e-05,
+      "loss": 0.5799,
+      "step": 6503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069964945807468e-05,
+      "loss": 0.5707,
+      "step": 6504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069303048987136e-05,
+      "loss": 0.5458,
+      "step": 6505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068641050815906e-05,
+      "loss": 0.5482,
+      "step": 6506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067978951302093e-05,
+      "loss": 0.5229,
+      "step": 6507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067316750454013e-05,
+      "loss": 0.5907,
+      "step": 6508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806665444827998e-05,
+      "loss": 0.5435,
+      "step": 6509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065992044788315e-05,
+      "loss": 0.5724,
+      "step": 6510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065329539987343e-05,
+      "loss": 0.5465,
+      "step": 6511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064666933885375e-05,
+      "loss": 0.5239,
+      "step": 6512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806400422649074e-05,
+      "loss": 0.5218,
+      "step": 6513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8063341417811763e-05,
+      "loss": 0.5665,
+      "step": 6514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062678507856763e-05,
+      "loss": 0.552,
+      "step": 6515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806201549663407e-05,
+      "loss": 0.5846,
+      "step": 6516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061352384152015e-05,
+      "loss": 0.5748,
+      "step": 6517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060689170418917e-05,
+      "loss": 0.5321,
+      "step": 6518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060025855443115e-05,
+      "loss": 0.596,
+      "step": 6519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059362439232938e-05,
+      "loss": 0.5386,
+      "step": 6520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058698921796712e-05,
+      "loss": 0.5798,
+      "step": 6521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805803530314278e-05,
+      "loss": 0.5298,
+      "step": 6522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057371583279473e-05,
+      "loss": 0.5117,
+      "step": 6523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056707762215125e-05,
+      "loss": 0.5832,
+      "step": 6524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805604383995808e-05,
+      "loss": 0.548,
+      "step": 6525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805537981651667e-05,
+      "loss": 0.6396,
+      "step": 6526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054715691899242e-05,
+      "loss": 0.5585,
+      "step": 6527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054051466114133e-05,
+      "loss": 0.5841,
+      "step": 6528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053387139169682e-05,
+      "loss": 0.5465,
+      "step": 6529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052722711074243e-05,
+      "loss": 0.5609,
+      "step": 6530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052058181836152e-05,
+      "loss": 0.6304,
+      "step": 6531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805139355146376e-05,
+      "loss": 0.5597,
+      "step": 6532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050728819965416e-05,
+      "loss": 0.5873,
+      "step": 6533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050063987349468e-05,
+      "loss": 0.5687,
+      "step": 6534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804939905362426e-05,
+      "loss": 0.5256,
+      "step": 6535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048734018798157e-05,
+      "loss": 0.5289,
+      "step": 6536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048068882879497e-05,
+      "loss": 0.5402,
+      "step": 6537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047403645876643e-05,
+      "loss": 0.5387,
+      "step": 6538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804673830779795e-05,
+      "loss": 0.5118,
+      "step": 6539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046072868651773e-05,
+      "loss": 0.5931,
+      "step": 6540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804540732844647e-05,
+      "loss": 0.5668,
+      "step": 6541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044741687190396e-05,
+      "loss": 0.5746,
+      "step": 6542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044075944891925e-05,
+      "loss": 0.62,
+      "step": 6543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80434101015594e-05,
+      "loss": 0.6095,
+      "step": 6544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80427441572012e-05,
+      "loss": 0.5425,
+      "step": 6545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804207811182568e-05,
+      "loss": 0.5883,
+      "step": 6546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804141196544121e-05,
+      "loss": 0.6131,
+      "step": 6547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040745718056153e-05,
+      "loss": 0.5327,
+      "step": 6548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040079369678883e-05,
+      "loss": 0.5778,
+      "step": 6549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039412920317762e-05,
+      "loss": 0.5946,
+      "step": 6550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803874636998117e-05,
+      "loss": 0.5161,
+      "step": 6551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803807971867747e-05,
+      "loss": 0.529,
+      "step": 6552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037412966415035e-05,
+      "loss": 0.5295,
+      "step": 6553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803674611320225e-05,
+      "loss": 0.5465,
+      "step": 6554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036079159047476e-05,
+      "loss": 0.5551,
+      "step": 6555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80354121039591e-05,
+      "loss": 0.5672,
+      "step": 6556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034744947945504e-05,
+      "loss": 0.5477,
+      "step": 6557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034077691015056e-05,
+      "loss": 0.6023,
+      "step": 6558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033410333176144e-05,
+      "loss": 0.5518,
+      "step": 6559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803274287443715e-05,
+      "loss": 0.5602,
+      "step": 6560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803207531480645e-05,
+      "loss": 0.5636,
+      "step": 6561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031407654292437e-05,
+      "loss": 0.6025,
+      "step": 6562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030739892903498e-05,
+      "loss": 0.5407,
+      "step": 6563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030072030648012e-05,
+      "loss": 0.5464,
+      "step": 6564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8029404067534372e-05,
+      "loss": 0.5777,
+      "step": 6565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028736003570966e-05,
+      "loss": 0.6238,
+      "step": 6566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028067838766188e-05,
+      "loss": 0.5719,
+      "step": 6567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802739957312843e-05,
+      "loss": 0.5754,
+      "step": 6568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802673120666608e-05,
+      "loss": 0.551,
+      "step": 6569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802606273938754e-05,
+      "loss": 0.496,
+      "step": 6570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025394171301198e-05,
+      "loss": 0.5849,
+      "step": 6571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802472550241546e-05,
+      "loss": 0.65,
+      "step": 6572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024056732738718e-05,
+      "loss": 0.6222,
+      "step": 6573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023387862279373e-05,
+      "loss": 0.5315,
+      "step": 6574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802271889104583e-05,
+      "loss": 0.5507,
+      "step": 6575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022049819046486e-05,
+      "loss": 0.5516,
+      "step": 6576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802138064628975e-05,
+      "loss": 0.5391,
+      "step": 6577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020711372784023e-05,
+      "loss": 0.5501,
+      "step": 6578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020041998537708e-05,
+      "loss": 0.6073,
+      "step": 6579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019372523559222e-05,
+      "loss": 0.5003,
+      "step": 6580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8018702947856966e-05,
+      "loss": 0.5776,
+      "step": 6581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801803327143935e-05,
+      "loss": 0.6184,
+      "step": 6582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017363494314788e-05,
+      "loss": 0.6038,
+      "step": 6583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801669361649169e-05,
+      "loss": 0.6076,
+      "step": 6584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016023637978475e-05,
+      "loss": 0.5171,
+      "step": 6585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801535355878355e-05,
+      "loss": 0.534,
+      "step": 6586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801468337891534e-05,
+      "loss": 0.5511,
+      "step": 6587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014013098382256e-05,
+      "loss": 0.588,
+      "step": 6588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013342717192718e-05,
+      "loss": 0.63,
+      "step": 6589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012672235355143e-05,
+      "loss": 0.6231,
+      "step": 6590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801200165287796e-05,
+      "loss": 0.6217,
+      "step": 6591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801133096976959e-05,
+      "loss": 0.5452,
+      "step": 6592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801066018603845e-05,
+      "loss": 0.5304,
+      "step": 6593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800998930169297e-05,
+      "loss": 0.5164,
+      "step": 6594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009318316741577e-05,
+      "loss": 0.5917,
+      "step": 6595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008647231192697e-05,
+      "loss": 0.5464,
+      "step": 6596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800797604505476e-05,
+      "loss": 0.5457,
+      "step": 6597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8007304758336194e-05,
+      "loss": 0.6253,
+      "step": 6598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8006633371045434e-05,
+      "loss": 0.5918,
+      "step": 6599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005961883190907e-05,
+      "loss": 0.5338,
+      "step": 6600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005290294781053e-05,
+      "loss": 0.5301,
+      "step": 6601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8004618605824303e-05,
+      "loss": 0.6339,
+      "step": 6602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003946816329095e-05,
+      "loss": 0.5598,
+      "step": 6603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800327492630387e-05,
+      "loss": 0.5819,
+      "step": 6604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800260293575706e-05,
+      "loss": 0.5357,
+      "step": 6605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800193084469711e-05,
+      "loss": 0.6378,
+      "step": 6606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800125865313246e-05,
+      "loss": 0.6028,
+      "step": 6607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000586361071555e-05,
+      "loss": 0.5747,
+      "step": 6608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999913968522835e-05,
+      "loss": 0.6568,
+      "step": 6609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799924147549475e-05,
+      "loss": 0.5725,
+      "step": 6610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998568881995743e-05,
+      "loss": 0.4952,
+      "step": 6611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997896188034262e-05,
+      "loss": 0.5478,
+      "step": 6612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997223393618756e-05,
+      "loss": 0.5604,
+      "step": 6613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799655049875768e-05,
+      "loss": 0.5681,
+      "step": 6614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799587750345948e-05,
+      "loss": 0.6053,
+      "step": 6615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799520440773261e-05,
+      "loss": 0.531,
+      "step": 6616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994531211585523e-05,
+      "loss": 0.5879,
+      "step": 6617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799385791502668e-05,
+      "loss": 0.5361,
+      "step": 6618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993184518064533e-05,
+      "loss": 0.5166,
+      "step": 6619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799251102070754e-05,
+      "loss": 0.5308,
+      "step": 6620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799183742296416e-05,
+      "loss": 0.5409,
+      "step": 6621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991163724842854e-05,
+      "loss": 0.5386,
+      "step": 6622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990489926352087e-05,
+      "loss": 0.6833,
+      "step": 6623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989816027500316e-05,
+      "loss": 0.5361,
+      "step": 6624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798914202829601e-05,
+      "loss": 0.5791,
+      "step": 6625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988467928747634e-05,
+      "loss": 0.5521,
+      "step": 6626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798779372886365e-05,
+      "loss": 0.5264,
+      "step": 6627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798711942865253e-05,
+      "loss": 0.6068,
+      "step": 6628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7986445028122745e-05,
+      "loss": 0.5131,
+      "step": 6629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798577052728276e-05,
+      "loss": 0.5579,
+      "step": 6630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985095926141055e-05,
+      "loss": 0.5365,
+      "step": 6631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984421224706095e-05,
+      "loss": 0.5638,
+      "step": 6632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983746422986357e-05,
+      "loss": 0.5516,
+      "step": 6633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983071520990316e-05,
+      "loss": 0.642,
+      "step": 6634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798239651872645e-05,
+      "loss": 0.5519,
+      "step": 6635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981721416203237e-05,
+      "loss": 0.5519,
+      "step": 6636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981046213429157e-05,
+      "loss": 0.5928,
+      "step": 6637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798037091041269e-05,
+      "loss": 0.4995,
+      "step": 6638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979695507162315e-05,
+      "loss": 0.6319,
+      "step": 6639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979020003686518e-05,
+      "loss": 0.5799,
+      "step": 6640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978344399993783e-05,
+      "loss": 0.6648,
+      "step": 6641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977668696092596e-05,
+      "loss": 0.5389,
+      "step": 6642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976992891991446e-05,
+      "loss": 0.6008,
+      "step": 6643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976316987698814e-05,
+      "loss": 0.5512,
+      "step": 6644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975640983223198e-05,
+      "loss": 0.5733,
+      "step": 6645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797496487857308e-05,
+      "loss": 0.5503,
+      "step": 6646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974288673756958e-05,
+      "loss": 0.5995,
+      "step": 6647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973612368783325e-05,
+      "loss": 0.53,
+      "step": 6648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972935963660674e-05,
+      "loss": 0.5584,
+      "step": 6649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79722594583975e-05,
+      "loss": 0.578,
+      "step": 6650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971582853002303e-05,
+      "loss": 0.5847,
+      "step": 6651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970906147483575e-05,
+      "loss": 0.4842,
+      "step": 6652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797022934184982e-05,
+      "loss": 0.5413,
+      "step": 6653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796955243610954e-05,
+      "loss": 0.535,
+      "step": 6654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968875430271234e-05,
+      "loss": 0.5957,
+      "step": 6655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796819832434341e-05,
+      "loss": 0.5791,
+      "step": 6656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7967521118334563e-05,
+      "loss": 0.6007,
+      "step": 6657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796684381225321e-05,
+      "loss": 0.5559,
+      "step": 6658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966166406107847e-05,
+      "loss": 0.6003,
+      "step": 6659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796548889990699e-05,
+      "loss": 0.5542,
+      "step": 6660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964811293659145e-05,
+      "loss": 0.5838,
+      "step": 6661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796413358737283e-05,
+      "loss": 0.5735,
+      "step": 6662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7963455781056543e-05,
+      "loss": 0.5227,
+      "step": 6663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796277787471881e-05,
+      "loss": 0.6363,
+      "step": 6664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796209986836814e-05,
+      "loss": 0.5564,
+      "step": 6665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796142176201305e-05,
+      "loss": 0.5742,
+      "step": 6666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960743555662057e-05,
+      "loss": 0.5609,
+      "step": 6667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796006524932368e-05,
+      "loss": 0.5449,
+      "step": 6668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7959386843006435e-05,
+      "loss": 0.5886,
+      "step": 6669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795870833671885e-05,
+      "loss": 0.5752,
+      "step": 6670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958029730469436e-05,
+      "loss": 0.5476,
+      "step": 6671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7957351024266726e-05,
+      "loss": 0.5855,
+      "step": 6672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795667221811924e-05,
+      "loss": 0.5859,
+      "step": 6673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955993312035513e-05,
+      "loss": 0.5432,
+      "step": 6674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955314306024055e-05,
+      "loss": 0.6207,
+      "step": 6675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795463520009341e-05,
+      "loss": 0.5083,
+      "step": 6676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7953955994252096e-05,
+      "loss": 0.5303,
+      "step": 6677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795327668850865e-05,
+      "loss": 0.5954,
+      "step": 6678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952597282871606e-05,
+      "loss": 0.563,
+      "step": 6679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951917777349495e-05,
+      "loss": 0.5776,
+      "step": 6680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951238171950852e-05,
+      "loss": 0.5713,
+      "step": 6681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795055846668421e-05,
+      "loss": 0.5146,
+      "step": 6682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949878661558108e-05,
+      "loss": 0.5418,
+      "step": 6683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949198756581086e-05,
+      "loss": 0.6119,
+      "step": 6684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948518751761682e-05,
+      "loss": 0.5874,
+      "step": 6685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947838647108437e-05,
+      "loss": 0.6152,
+      "step": 6686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947158442629896e-05,
+      "loss": 0.5131,
+      "step": 6687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946478138334597e-05,
+      "loss": 0.5923,
+      "step": 6688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945797734231088e-05,
+      "loss": 0.5622,
+      "step": 6689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945117230327915e-05,
+      "loss": 0.6016,
+      "step": 6690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944436626633625e-05,
+      "loss": 0.5629,
+      "step": 6691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943755923156763e-05,
+      "loss": 0.603,
+      "step": 6692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943075119905885e-05,
+      "loss": 0.5228,
+      "step": 6693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794239421688954e-05,
+      "loss": 0.6638,
+      "step": 6694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941713214116275e-05,
+      "loss": 0.6106,
+      "step": 6695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794103211159465e-05,
+      "loss": 0.5764,
+      "step": 6696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794035090933321e-05,
+      "loss": 0.6213,
+      "step": 6697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939669607340525e-05,
+      "loss": 0.589,
+      "step": 6698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7938988205625145e-05,
+      "loss": 0.5891,
+      "step": 6699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7938306704195627e-05,
+      "loss": 0.5769,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937625103060536e-05,
+      "loss": 0.5036,
+      "step": 6701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936943402228423e-05,
+      "loss": 0.5698,
+      "step": 6702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936261601707858e-05,
+      "loss": 0.6035,
+      "step": 6703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935579701507406e-05,
+      "loss": 0.6043,
+      "step": 6704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934897701635624e-05,
+      "loss": 0.5786,
+      "step": 6705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934215602101085e-05,
+      "loss": 0.633,
+      "step": 6706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.5878,
+      "step": 6707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932851104077996e-05,
+      "loss": 0.574,
+      "step": 6708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793216870560659e-05,
+      "loss": 0.5517,
+      "step": 6709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931486207506695e-05,
+      "loss": 0.5868,
+      "step": 6710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930803609786892e-05,
+      "loss": 0.5024,
+      "step": 6711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930120912455747e-05,
+      "loss": 0.5775,
+      "step": 6712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929438115521845e-05,
+      "loss": 0.5722,
+      "step": 6713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928755218993755e-05,
+      "loss": 0.5675,
+      "step": 6714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928072222880053e-05,
+      "loss": 0.5555,
+      "step": 6715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792738912718932e-05,
+      "loss": 0.5395,
+      "step": 6716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7926705931930137e-05,
+      "loss": 0.567,
+      "step": 6717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7926022637111086e-05,
+      "loss": 0.571,
+      "step": 6718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925339242740743e-05,
+      "loss": 0.5024,
+      "step": 6719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924655748827695e-05,
+      "loss": 0.5594,
+      "step": 6720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792397215538053e-05,
+      "loss": 0.6271,
+      "step": 6721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923288462407828e-05,
+      "loss": 0.5115,
+      "step": 6722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792260466991818e-05,
+      "loss": 0.5911,
+      "step": 6723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921920777920175e-05,
+      "loss": 0.5496,
+      "step": 6724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79212367864224e-05,
+      "loss": 0.5476,
+      "step": 6725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7920552695433448e-05,
+      "loss": 0.562,
+      "step": 6726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791986850496191e-05,
+      "loss": 0.5685,
+      "step": 6727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919184215016382e-05,
+      "loss": 0.5,
+      "step": 6728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918499825605455e-05,
+      "loss": 0.5832,
+      "step": 6729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791781533673773e-05,
+      "loss": 0.5261,
+      "step": 6730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917130748421795e-05,
+      "loss": 0.5489,
+      "step": 6731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7916446060666254e-05,
+      "loss": 0.6143,
+      "step": 6732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791576127347971e-05,
+      "loss": 0.6109,
+      "step": 6733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915076386870764e-05,
+      "loss": 0.5844,
+      "step": 6734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914391400848013e-05,
+      "loss": 0.5277,
+      "step": 6735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791370631542006e-05,
+      "loss": 0.5249,
+      "step": 6736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913021130595516e-05,
+      "loss": 0.5481,
+      "step": 6737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912335846382982e-05,
+      "loss": 0.5407,
+      "step": 6738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911650462791067e-05,
+      "loss": 0.5417,
+      "step": 6739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910964979828377e-05,
+      "loss": 0.6256,
+      "step": 6740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910279397503525e-05,
+      "loss": 0.6571,
+      "step": 6741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909593715825124e-05,
+      "loss": 0.6047,
+      "step": 6742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790890793480178e-05,
+      "loss": 0.6039,
+      "step": 6743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790822205444211e-05,
+      "loss": 0.5266,
+      "step": 6744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907536074754726e-05,
+      "loss": 0.5323,
+      "step": 6745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790684999574825e-05,
+      "loss": 0.5751,
+      "step": 6746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906163817431294e-05,
+      "loss": 0.6419,
+      "step": 6747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905477539812475e-05,
+      "loss": 0.5603,
+      "step": 6748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904791162900418e-05,
+      "loss": 0.5957,
+      "step": 6749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904104686703737e-05,
+      "loss": 0.6001,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903418111231064e-05,
+      "loss": 0.6017,
+      "step": 6751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902731436491015e-05,
+      "loss": 0.545,
+      "step": 6752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902044662492216e-05,
+      "loss": 0.6206,
+      "step": 6753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901357789243294e-05,
+      "loss": 0.6342,
+      "step": 6754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900670816752875e-05,
+      "loss": 0.618,
+      "step": 6755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789998374502959e-05,
+      "loss": 0.6095,
+      "step": 6756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789929657408206e-05,
+      "loss": 0.5381,
+      "step": 6757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789860930391893e-05,
+      "loss": 0.5206,
+      "step": 6758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897921934548823e-05,
+      "loss": 0.5635,
+      "step": 6759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789723446598037e-05,
+      "loss": 0.5984,
+      "step": 6760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896546898222218e-05,
+      "loss": 0.5428,
+      "step": 6761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789585923128299e-05,
+      "loss": 0.6388,
+      "step": 6762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789517146517133e-05,
+      "loss": 0.5657,
+      "step": 6763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789448359989587e-05,
+      "loss": 0.5231,
+      "step": 6764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893795635465257e-05,
+      "loss": 0.5656,
+      "step": 6765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893107571888127e-05,
+      "loss": 0.5424,
+      "step": 6766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892419409173123e-05,
+      "loss": 0.5715,
+      "step": 6767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789173114732889e-05,
+      "loss": 0.5436,
+      "step": 6768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7891042786364074e-05,
+      "loss": 0.5918,
+      "step": 6769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890354326287318e-05,
+      "loss": 0.5135,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889665767107267e-05,
+      "loss": 0.5153,
+      "step": 6771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888977108832578e-05,
+      "loss": 0.6072,
+      "step": 6772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788828835147189e-05,
+      "loss": 0.6127,
+      "step": 6773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788759949503386e-05,
+      "loss": 0.5753,
+      "step": 6774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886910539527137e-05,
+      "loss": 0.5399,
+      "step": 6775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886221484960375e-05,
+      "loss": 0.5736,
+      "step": 6776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885532331342234e-05,
+      "loss": 0.5747,
+      "step": 6777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788484307868136e-05,
+      "loss": 0.643,
+      "step": 6778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788415372698642e-05,
+      "loss": 0.6167,
+      "step": 6779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883464276266064e-05,
+      "loss": 0.5716,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882774726528954e-05,
+      "loss": 0.5484,
+      "step": 6781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788208507778376e-05,
+      "loss": 0.5545,
+      "step": 6782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881395330039128e-05,
+      "loss": 0.5564,
+      "step": 6783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880705483303732e-05,
+      "loss": 0.5636,
+      "step": 6784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880015537586234e-05,
+      "loss": 0.5933,
+      "step": 6785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879325492895298e-05,
+      "loss": 0.577,
+      "step": 6786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7878635349239593e-05,
+      "loss": 0.5903,
+      "step": 6787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877945106627787e-05,
+      "loss": 0.6184,
+      "step": 6788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787725476506855e-05,
+      "loss": 0.6146,
+      "step": 6789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7876564324570553e-05,
+      "loss": 0.5518,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7875873785142464e-05,
+      "loss": 0.58,
+      "step": 6791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7875183146792963e-05,
+      "loss": 0.4813,
+      "step": 6792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787449240953072e-05,
+      "loss": 0.587,
+      "step": 6793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873801573364406e-05,
+      "loss": 0.5603,
+      "step": 6794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873110638302708e-05,
+      "loss": 0.6148,
+      "step": 6795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78724196043543e-05,
+      "loss": 0.5595,
+      "step": 6796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787172847152786e-05,
+      "loss": 0.5379,
+      "step": 6797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871037239832066e-05,
+      "loss": 0.6281,
+      "step": 6798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870345909275605e-05,
+      "loss": 0.521,
+      "step": 6799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7869654479867163e-05,
+      "loss": 0.5333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868962951615415e-05,
+      "loss": 0.6011,
+      "step": 6801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868271324529053e-05,
+      "loss": 0.5374,
+      "step": 6802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867579598616765e-05,
+      "loss": 0.6136,
+      "step": 6803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866887773887235e-05,
+      "loss": 0.5407,
+      "step": 6804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866195850349154e-05,
+      "loss": 0.5682,
+      "step": 6805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786550382801121e-05,
+      "loss": 0.5043,
+      "step": 6806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78648117068821e-05,
+      "loss": 0.5931,
+      "step": 6807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864119486970513e-05,
+      "loss": 0.5703,
+      "step": 6808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863427168285148e-05,
+      "loss": 0.4833,
+      "step": 6809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862734750834694e-05,
+      "loss": 0.554,
+      "step": 6810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862042234627853e-05,
+      "loss": 0.55,
+      "step": 6811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861349619673318e-05,
+      "loss": 0.5092,
+      "step": 6812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860656905979796e-05,
+      "loss": 0.5569,
+      "step": 6813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7859964093555978e-05,
+      "loss": 0.5142,
+      "step": 6814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7859271182410576e-05,
+      "loss": 0.5055,
+      "step": 6815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858578172552285e-05,
+      "loss": 0.6107,
+      "step": 6816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785788506398981e-05,
+      "loss": 0.5916,
+      "step": 6817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857191856731865e-05,
+      "loss": 0.5642,
+      "step": 6818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856498550787144e-05,
+      "loss": 0.5008,
+      "step": 6819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855805146164364e-05,
+      "loss": 0.5327,
+      "step": 6820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785511164287223e-05,
+      "loss": 0.5135,
+      "step": 6821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854418040919456e-05,
+      "loss": 0.5823,
+      "step": 6822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853724340314752e-05,
+      "loss": 0.5882,
+      "step": 6823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853030541066827e-05,
+      "loss": 0.5275,
+      "step": 6824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78523366431844e-05,
+      "loss": 0.5765,
+      "step": 6825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7851642646676186e-05,
+      "loss": 0.4792,
+      "step": 6826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78509485515509e-05,
+      "loss": 0.5366,
+      "step": 6827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785025435781726e-05,
+      "loss": 0.6043,
+      "step": 6828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849560065483985e-05,
+      "loss": 0.6318,
+      "step": 6829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78488656745598e-05,
+      "loss": 0.6223,
+      "step": 6830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7848171185053416e-05,
+      "loss": 0.5971,
+      "step": 6831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847476596973567e-05,
+      "loss": 0.5166,
+      "step": 6832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846781910328973e-05,
+      "loss": 0.5123,
+      "step": 6833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846087125128355e-05,
+      "loss": 0.5513,
+      "step": 6834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845392241380444e-05,
+      "loss": 0.5589,
+      "step": 6835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784469725909397e-05,
+      "loss": 0.5256,
+      "step": 6836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844002178277656e-05,
+      "loss": 0.5439,
+      "step": 6837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843306998940238e-05,
+      "loss": 0.5154,
+      "step": 6838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7842611721090445e-05,
+      "loss": 0.51,
+      "step": 6839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841916344737005e-05,
+      "loss": 0.5658,
+      "step": 6840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784122086988866e-05,
+      "loss": 0.5074,
+      "step": 6841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784052529655414e-05,
+      "loss": 0.5413,
+      "step": 6842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839829624742185e-05,
+      "loss": 0.5314,
+      "step": 6843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783913385446153e-05,
+      "loss": 0.5708,
+      "step": 6844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7838437985720918e-05,
+      "loss": 0.4839,
+      "step": 6845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783774201852908e-05,
+      "loss": 0.5447,
+      "step": 6846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7837045952894768e-05,
+      "loss": 0.5814,
+      "step": 6847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7836349788826718e-05,
+      "loss": 0.5383,
+      "step": 6848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835653526333674e-05,
+      "loss": 0.6247,
+      "step": 6849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834957165424386e-05,
+      "loss": 0.5405,
+      "step": 6850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834260706107597e-05,
+      "loss": 0.5088,
+      "step": 6851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833564148392054e-05,
+      "loss": 0.5934,
+      "step": 6852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832867492286506e-05,
+      "loss": 0.5891,
+      "step": 6853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832170737799703e-05,
+      "loss": 0.5558,
+      "step": 6854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78314738849404e-05,
+      "loss": 0.5072,
+      "step": 6855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830776933717343e-05,
+      "loss": 0.5696,
+      "step": 6856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830079884139292e-05,
+      "loss": 0.5503,
+      "step": 6857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829382736214996e-05,
+      "loss": 0.5639,
+      "step": 6858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7828685489953215e-05,
+      "loss": 0.5669,
+      "step": 6859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827988145362705e-05,
+      "loss": 0.5637,
+      "step": 6860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782729070245223e-05,
+      "loss": 0.5875,
+      "step": 6861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782659316123054e-05,
+      "loss": 0.5516,
+      "step": 6862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825895521706404e-05,
+      "loss": 0.5233,
+      "step": 6863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782519778388858e-05,
+      "loss": 0.5944,
+      "step": 6864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782449994778584e-05,
+      "loss": 0.5291,
+      "step": 6865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823802013406935e-05,
+      "loss": 0.4932,
+      "step": 6866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823103980760642e-05,
+      "loss": 0.5895,
+      "step": 6867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7822405849855723e-05,
+      "loss": 0.5234,
+      "step": 6868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782170762070095e-05,
+      "loss": 0.5272,
+      "step": 6869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821009293305093e-05,
+      "loss": 0.5132,
+      "step": 6870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782031086767692e-05,
+      "loss": 0.5351,
+      "step": 6871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819612343825202e-05,
+      "loss": 0.6169,
+      "step": 6872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818913721758718e-05,
+      "loss": 0.4979,
+      "step": 6873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818215001486237e-05,
+      "loss": 0.6019,
+      "step": 6874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781751618301654e-05,
+      "loss": 0.5481,
+      "step": 6875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78168172663584e-05,
+      "loss": 0.542,
+      "step": 6876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78161182515206e-05,
+      "loss": 0.6517,
+      "step": 6877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815419138511917e-05,
+      "loss": 0.5496,
+      "step": 6878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814719927341126e-05,
+      "loss": 0.5412,
+      "step": 6879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781402061801702e-05,
+      "loss": 0.4878,
+      "step": 6880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813321210548377e-05,
+      "loss": 0.6146,
+      "step": 6881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781262170494398e-05,
+      "loss": 0.5574,
+      "step": 6882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811922101212622e-05,
+      "loss": 0.5098,
+      "step": 6883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811222399363082e-05,
+      "loss": 0.6083,
+      "step": 6884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781052259940415e-05,
+      "loss": 0.5311,
+      "step": 6885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780982270134462e-05,
+      "loss": 0.5578,
+      "step": 6886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7809122705193274e-05,
+      "loss": 0.5366,
+      "step": 6887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808422610958913e-05,
+      "loss": 0.5838,
+      "step": 6888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807722418650326e-05,
+      "loss": 0.5768,
+      "step": 6889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807022128276307e-05,
+      "loss": 0.6002,
+      "step": 6890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806321739845654e-05,
+      "loss": 0.5236,
+      "step": 6891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805621253367164e-05,
+      "loss": 0.5656,
+      "step": 6892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7804920668849634e-05,
+      "loss": 0.624,
+      "step": 6893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780421998630186e-05,
+      "loss": 0.5066,
+      "step": 6894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780351920573265e-05,
+      "loss": 0.5878,
+      "step": 6895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802818327150797e-05,
+      "loss": 0.599,
+      "step": 6896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802117350565114e-05,
+      "loss": 0.5961,
+      "step": 6897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801416275984394e-05,
+      "loss": 0.564,
+      "step": 6898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800715103417452e-05,
+      "loss": 0.5907,
+      "step": 6899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800013832873095e-05,
+      "loss": 0.5385,
+      "step": 6900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799312464360125e-05,
+      "loss": 0.5501,
+      "step": 6901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779861099788735e-05,
+      "loss": 0.5814,
+      "step": 6902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797909433463587e-05,
+      "loss": 0.5257,
+      "step": 6903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797207771097645e-05,
+      "loss": 0.5847,
+      "step": 6904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796506010798336e-05,
+      "loss": 0.5373,
+      "step": 6905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795804152574474e-05,
+      "loss": 0.5538,
+      "step": 6906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795102196434876e-05,
+      "loss": 0.5452,
+      "step": 6907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794400142388356e-05,
+      "loss": 0.5664,
+      "step": 6908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779369799044374e-05,
+      "loss": 0.6044,
+      "step": 6909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792995740609834e-05,
+      "loss": 0.5931,
+      "step": 6910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792293392895467e-05,
+      "loss": 0.5512,
+      "step": 6911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779159094730946e-05,
+      "loss": 0.5684,
+      "step": 6912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790888403860634e-05,
+      "loss": 0.528,
+      "step": 6913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790185762557813e-05,
+      "loss": 0.5746,
+      "step": 6914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7789483023409822e-05,
+      "loss": 0.5878,
+      "step": 6915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788780186425487e-05,
+      "loss": 0.5925,
+      "step": 6916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778807725161364e-05,
+      "loss": 0.561,
+      "step": 6917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787374218983103e-05,
+      "loss": 0.5388,
+      "step": 6918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778667108854271e-05,
+      "loss": 0.5779,
+      "step": 6919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785967860301293e-05,
+      "loss": 0.6286,
+      "step": 6920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785264534267687e-05,
+      "loss": 0.5836,
+      "step": 6921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784561110450717e-05,
+      "loss": 0.5462,
+      "step": 6922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783857588859226e-05,
+      "loss": 0.5535,
+      "step": 6923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778315396950205e-05,
+      "loss": 0.4984,
+      "step": 6924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778245025238802e-05,
+      "loss": 0.5188,
+      "step": 6925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778174643752598e-05,
+      "loss": 0.6041,
+      "step": 6926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778104252492477e-05,
+      "loss": 0.5887,
+      "step": 6927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778033851459323e-05,
+      "loss": 0.5863,
+      "step": 6928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77796344065402e-05,
+      "loss": 0.5825,
+      "step": 6929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777893020077453e-05,
+      "loss": 0.5314,
+      "step": 6930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778225897305058e-05,
+      "loss": 0.4637,
+      "step": 6931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777521496140636e-05,
+      "loss": 0.5612,
+      "step": 6932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776816997290107e-05,
+      "loss": 0.5593,
+      "step": 6933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777611240076232e-05,
+      "loss": 0.5657,
+      "step": 6934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775407706566125e-05,
+      "loss": 0.6184,
+      "step": 6935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774702914710375e-05,
+      "loss": 0.5607,
+      "step": 6936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773998025203924e-05,
+      "loss": 0.4863,
+      "step": 6937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773293038055617e-05,
+      "loss": 0.5593,
+      "step": 6938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777258795327432e-05,
+      "loss": 0.5928,
+      "step": 6939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771882770868875e-05,
+      "loss": 0.5658,
+      "step": 6940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771177490848153e-05,
+      "loss": 0.5255,
+      "step": 6941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770472113221006e-05,
+      "loss": 0.5407,
+      "step": 6942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769766637996296e-05,
+      "loss": 0.5607,
+      "step": 6943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769061065182874e-05,
+      "loss": 0.5644,
+      "step": 6944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768355394789617e-05,
+      "loss": 0.5969,
+      "step": 6945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776764962682538e-05,
+      "loss": 0.5553,
+      "step": 6946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776694376129903e-05,
+      "loss": 0.5711,
+      "step": 6947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776623779821943e-05,
+      "loss": 0.6043,
+      "step": 6948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765531737595446e-05,
+      "loss": 0.5176,
+      "step": 6949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776482557943595e-05,
+      "loss": 0.5915,
+      "step": 6950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776411932374981e-05,
+      "loss": 0.5788,
+      "step": 6951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763412970545895e-05,
+      "loss": 0.5822,
+      "step": 6952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776270651983308e-05,
+      "loss": 0.566,
+      "step": 6953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761999971620236e-05,
+      "loss": 0.5528,
+      "step": 6954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761293325916238e-05,
+      "loss": 0.5709,
+      "step": 6955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760586582729962e-05,
+      "loss": 0.6158,
+      "step": 6956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775987974207028e-05,
+      "loss": 0.5526,
+      "step": 6957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759172803946076e-05,
+      "loss": 0.6275,
+      "step": 6958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775846576836623e-05,
+      "loss": 0.5631,
+      "step": 6959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757758635339613e-05,
+      "loss": 0.6566,
+      "step": 6960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757051404875117e-05,
+      "loss": 0.6135,
+      "step": 6961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7756344076981622e-05,
+      "loss": 0.5446,
+      "step": 6962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775563665166801e-05,
+      "loss": 0.5647,
+      "step": 6963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775492912894317e-05,
+      "loss": 0.5607,
+      "step": 6964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7754221508815982e-05,
+      "loss": 0.5694,
+      "step": 6965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7753513791295338e-05,
+      "loss": 0.5421,
+      "step": 6966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752805976390126e-05,
+      "loss": 0.569,
+      "step": 6967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752098064109237e-05,
+      "loss": 0.5068,
+      "step": 6968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7751390054461562e-05,
+      "loss": 0.5641,
+      "step": 6969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7750681947456e-05,
+      "loss": 0.5983,
+      "step": 6970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774997374310143e-05,
+      "loss": 0.6278,
+      "step": 6971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774926544140676e-05,
+      "loss": 0.584,
+      "step": 6972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774855704238088e-05,
+      "loss": 0.5773,
+      "step": 6973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747848546032692e-05,
+      "loss": 0.6138,
+      "step": 6974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774713995237109e-05,
+      "loss": 0.5004,
+      "step": 6975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774643126140498e-05,
+      "loss": 0.6179,
+      "step": 6976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745722473143255e-05,
+      "loss": 0.5627,
+      "step": 6977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745013587594825e-05,
+      "loss": 0.6292,
+      "step": 6978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744304604768588e-05,
+      "loss": 0.5826,
+      "step": 6979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743595524673455e-05,
+      "loss": 0.5643,
+      "step": 6980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742886347318326e-05,
+      "loss": 0.5244,
+      "step": 6981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774217707271211e-05,
+      "loss": 0.5129,
+      "step": 6982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741467700863717e-05,
+      "loss": 0.5988,
+      "step": 6983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740758231782055e-05,
+      "loss": 0.5669,
+      "step": 6984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774004866547604e-05,
+      "loss": 0.5128,
+      "step": 6985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739339001954577e-05,
+      "loss": 0.5376,
+      "step": 6986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738629241226583e-05,
+      "loss": 0.5461,
+      "step": 6987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737919383300976e-05,
+      "loss": 0.5775,
+      "step": 6988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737209428186662e-05,
+      "loss": 0.4855,
+      "step": 6989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736499375892567e-05,
+      "loss": 0.5873,
+      "step": 6990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773578922642761e-05,
+      "loss": 0.5609,
+      "step": 6991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735078979800705e-05,
+      "loss": 0.5169,
+      "step": 6992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7734368636020774e-05,
+      "loss": 0.6229,
+      "step": 6993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733658195096737e-05,
+      "loss": 0.519,
+      "step": 6994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732947657037524e-05,
+      "loss": 0.5586,
+      "step": 6995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732237021852055e-05,
+      "loss": 0.561,
+      "step": 6996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7731526289549254e-05,
+      "loss": 0.5856,
+      "step": 6997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773081546013805e-05,
+      "loss": 0.5171,
+      "step": 6998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.5272,
+      "step": 6999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772939351002615e-05,
+      "loss": 0.5243,
+      "step": 7000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728682389343308e-05,
+      "loss": 0.5727,
+      "step": 7001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727971171587785e-05,
+      "loss": 0.6042,
+      "step": 7002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727259856768507e-05,
+      "loss": 0.6138,
+      "step": 7003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726548444894418e-05,
+      "loss": 0.6007,
+      "step": 7004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725836935974447e-05,
+      "loss": 0.5037,
+      "step": 7005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772512533001753e-05,
+      "loss": 0.6115,
+      "step": 7006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724413627032607e-05,
+      "loss": 0.5969,
+      "step": 7007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723701827028615e-05,
+      "loss": 0.5811,
+      "step": 7008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722989930014498e-05,
+      "loss": 0.5564,
+      "step": 7009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772227793599919e-05,
+      "loss": 0.5432,
+      "step": 7010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721565844991643e-05,
+      "loss": 0.5539,
+      "step": 7011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7720853657000797e-05,
+      "loss": 0.5286,
+      "step": 7012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7720141372035592e-05,
+      "loss": 0.5334,
+      "step": 7013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719428990104983e-05,
+      "loss": 0.5769,
+      "step": 7014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718716511217913e-05,
+      "loss": 0.5445,
+      "step": 7015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718003935383332e-05,
+      "loss": 0.5629,
+      "step": 7016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717291262610188e-05,
+      "loss": 0.5518,
+      "step": 7017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716578492907434e-05,
+      "loss": 0.6159,
+      "step": 7018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715865626284023e-05,
+      "loss": 0.5257,
+      "step": 7019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715152662748907e-05,
+      "loss": 0.5059,
+      "step": 7020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771443960231104e-05,
+      "loss": 0.5388,
+      "step": 7021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771372644497938e-05,
+      "loss": 0.5422,
+      "step": 7022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7713013190762886e-05,
+      "loss": 0.5741,
+      "step": 7023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771229983967051e-05,
+      "loss": 0.5447,
+      "step": 7024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711586391711217e-05,
+      "loss": 0.5581,
+      "step": 7025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710872846893973e-05,
+      "loss": 0.5956,
+      "step": 7026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710159205227724e-05,
+      "loss": 0.5455,
+      "step": 7027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709445466721452e-05,
+      "loss": 0.5444,
+      "step": 7028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708731631384112e-05,
+      "loss": 0.5734,
+      "step": 7029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770801769922467e-05,
+      "loss": 0.515,
+      "step": 7030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707303670252093e-05,
+      "loss": 0.6001,
+      "step": 7031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770658954447535e-05,
+      "loss": 0.5862,
+      "step": 7032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770587532190341e-05,
+      "loss": 0.5598,
+      "step": 7033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705161002545247e-05,
+      "loss": 0.5171,
+      "step": 7034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770444658640983e-05,
+      "loss": 0.5516,
+      "step": 7035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770373207350613e-05,
+      "loss": 0.52,
+      "step": 7036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703017463843126e-05,
+      "loss": 0.4945,
+      "step": 7037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770230275742979e-05,
+      "loss": 0.5386,
+      "step": 7038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701587954275102e-05,
+      "loss": 0.5966,
+      "step": 7039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700873054388034e-05,
+      "loss": 0.6201,
+      "step": 7040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700158057777573e-05,
+      "loss": 0.5862,
+      "step": 7041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699442964452693e-05,
+      "loss": 0.4971,
+      "step": 7042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769872777442238e-05,
+      "loss": 0.5928,
+      "step": 7043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698012487695613e-05,
+      "loss": 0.4798,
+      "step": 7044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769729710428138e-05,
+      "loss": 0.5655,
+      "step": 7045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696581624188663e-05,
+      "loss": 0.6406,
+      "step": 7046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769586604742645e-05,
+      "loss": 0.5518,
+      "step": 7047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7695150374003728e-05,
+      "loss": 0.5866,
+      "step": 7048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694434603929488e-05,
+      "loss": 0.6134,
+      "step": 7049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693718737212716e-05,
+      "loss": 0.5482,
+      "step": 7050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769300277386241e-05,
+      "loss": 0.5954,
+      "step": 7051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692286713887555e-05,
+      "loss": 0.5727,
+      "step": 7052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769157055729715e-05,
+      "loss": 0.6465,
+      "step": 7053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690854304100184e-05,
+      "loss": 0.5693,
+      "step": 7054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769013795430566e-05,
+      "loss": 0.4947,
+      "step": 7055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689421507922573e-05,
+      "loss": 0.5803,
+      "step": 7056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768870496495992e-05,
+      "loss": 0.5417,
+      "step": 7057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687988325426705e-05,
+      "loss": 0.5773,
+      "step": 7058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768727158933192e-05,
+      "loss": 0.5336,
+      "step": 7059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686554756684576e-05,
+      "loss": 0.5334,
+      "step": 7060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685837827493676e-05,
+      "loss": 0.5137,
+      "step": 7061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768512080176822e-05,
+      "loss": 0.5028,
+      "step": 7062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684403679517218e-05,
+      "loss": 0.6355,
+      "step": 7063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683686460749677e-05,
+      "loss": 0.5375,
+      "step": 7064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76829691454746e-05,
+      "loss": 0.4813,
+      "step": 7065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7682251733701004e-05,
+      "loss": 0.5831,
+      "step": 7066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681534225437893e-05,
+      "loss": 0.5962,
+      "step": 7067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680816620694284e-05,
+      "loss": 0.5861,
+      "step": 7068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680098919479186e-05,
+      "loss": 0.5677,
+      "step": 7069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679381121801617e-05,
+      "loss": 0.5642,
+      "step": 7070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678663227670587e-05,
+      "loss": 0.6114,
+      "step": 7071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677945237095122e-05,
+      "loss": 0.565,
+      "step": 7072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677227150084234e-05,
+      "loss": 0.5628,
+      "step": 7073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767650896664694e-05,
+      "loss": 0.5359,
+      "step": 7074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7675790686792264e-05,
+      "loss": 0.5485,
+      "step": 7075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7675072310529227e-05,
+      "loss": 0.4802,
+      "step": 7076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674353837866854e-05,
+      "loss": 0.5646,
+      "step": 7077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673635268814168e-05,
+      "loss": 0.5911,
+      "step": 7078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767291660338019e-05,
+      "loss": 0.545,
+      "step": 7079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672197841573954e-05,
+      "loss": 0.5322,
+      "step": 7080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767147898340448e-05,
+      "loss": 0.5114,
+      "step": 7081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76707600288808e-05,
+      "loss": 0.5259,
+      "step": 7082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670040978011947e-05,
+      "loss": 0.5949,
+      "step": 7083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669321830806948e-05,
+      "loss": 0.5669,
+      "step": 7084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766860258727484e-05,
+      "loss": 0.5764,
+      "step": 7085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667883247424647e-05,
+      "loss": 0.5029,
+      "step": 7086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667163811265417e-05,
+      "loss": 0.6064,
+      "step": 7087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666444278806178e-05,
+      "loss": 0.5037,
+      "step": 7088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665724650055972e-05,
+      "loss": 0.5584,
+      "step": 7089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665004925023833e-05,
+      "loss": 0.5557,
+      "step": 7090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76642851037188e-05,
+      "loss": 0.5777,
+      "step": 7091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766356518614992e-05,
+      "loss": 0.6607,
+      "step": 7092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662845172326234e-05,
+      "loss": 0.5939,
+      "step": 7093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766212506225678e-05,
+      "loss": 0.591,
+      "step": 7094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766140485595061e-05,
+      "loss": 0.5204,
+      "step": 7095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7660684553416763e-05,
+      "loss": 0.6065,
+      "step": 7096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659964154664288e-05,
+      "loss": 0.5461,
+      "step": 7097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659243659702236e-05,
+      "loss": 0.6146,
+      "step": 7098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7658523068539653e-05,
+      "loss": 0.5631,
+      "step": 7099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765780238118559e-05,
+      "loss": 0.5429,
+      "step": 7100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657081597649106e-05,
+      "loss": 0.5281,
+      "step": 7101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656360717939244e-05,
+      "loss": 0.5791,
+      "step": 7102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7655639742065064e-05,
+      "loss": 0.5378,
+      "step": 7103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765491867003562e-05,
+      "loss": 0.5934,
+      "step": 7104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654197501859967e-05,
+      "loss": 0.56,
+      "step": 7105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653476237547165e-05,
+      "loss": 0.5766,
+      "step": 7106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652754877106275e-05,
+      "loss": 0.6305,
+      "step": 7107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652033420546356e-05,
+      "loss": 0.5621,
+      "step": 7108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651311867876465e-05,
+      "loss": 0.6045,
+      "step": 7109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765059021910567e-05,
+      "loss": 0.5895,
+      "step": 7110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649868474243038e-05,
+      "loss": 0.5643,
+      "step": 7111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649146633297623e-05,
+      "loss": 0.4808,
+      "step": 7112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76484246962785e-05,
+      "loss": 0.5313,
+      "step": 7113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647702663194735e-05,
+      "loss": 0.6162,
+      "step": 7114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646980534055392e-05,
+      "loss": 0.5612,
+      "step": 7115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764625830886955e-05,
+      "loss": 0.5418,
+      "step": 7116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645535987646276e-05,
+      "loss": 0.6245,
+      "step": 7117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764481357039464e-05,
+      "loss": 0.6025,
+      "step": 7118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644091057123717e-05,
+      "loss": 0.5561,
+      "step": 7119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643368447842583e-05,
+      "loss": 0.5495,
+      "step": 7120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642645742560313e-05,
+      "loss": 0.6019,
+      "step": 7121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641922941285984e-05,
+      "loss": 0.6096,
+      "step": 7122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641200044028677e-05,
+      "loss": 0.5992,
+      "step": 7123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640477050797466e-05,
+      "loss": 0.5028,
+      "step": 7124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763975396160144e-05,
+      "loss": 0.5622,
+      "step": 7125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639030776449672e-05,
+      "loss": 0.5624,
+      "step": 7126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763830749535125e-05,
+      "loss": 0.5651,
+      "step": 7127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637584118315258e-05,
+      "loss": 0.4946,
+      "step": 7128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636860645350785e-05,
+      "loss": 0.613,
+      "step": 7129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763613707646691e-05,
+      "loss": 0.6168,
+      "step": 7130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763541341167273e-05,
+      "loss": 0.4851,
+      "step": 7131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634689650977325e-05,
+      "loss": 0.5773,
+      "step": 7132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633965794389797e-05,
+      "loss": 0.589,
+      "step": 7133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633241841919228e-05,
+      "loss": 0.5568,
+      "step": 7134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763251779357471e-05,
+      "loss": 0.5572,
+      "step": 7135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631793649365344e-05,
+      "loss": 0.5001,
+      "step": 7136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631069409300225e-05,
+      "loss": 0.5175,
+      "step": 7137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630345073388445e-05,
+      "loss": 0.6297,
+      "step": 7138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629620641639102e-05,
+      "loss": 0.5542,
+      "step": 7139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628896114061296e-05,
+      "loss": 0.5943,
+      "step": 7140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762817149066413e-05,
+      "loss": 0.5824,
+      "step": 7141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627446771456703e-05,
+      "loss": 0.5638,
+      "step": 7142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626721956448116e-05,
+      "loss": 0.5582,
+      "step": 7143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625997045647475e-05,
+      "loss": 0.5476,
+      "step": 7144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625272039063884e-05,
+      "loss": 0.5635,
+      "step": 7145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624546936706445e-05,
+      "loss": 0.5782,
+      "step": 7146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623821738584276e-05,
+      "loss": 0.5329,
+      "step": 7147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623096444706476e-05,
+      "loss": 0.5967,
+      "step": 7148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622371055082156e-05,
+      "loss": 0.606,
+      "step": 7149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621645569720432e-05,
+      "loss": 0.5619,
+      "step": 7150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620919988630413e-05,
+      "loss": 0.5659,
+      "step": 7151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620194311821208e-05,
+      "loss": 0.5337,
+      "step": 7152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761946853930194e-05,
+      "loss": 0.5705,
+      "step": 7153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618742671081713e-05,
+      "loss": 0.5229,
+      "step": 7154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618016707169658e-05,
+      "loss": 0.5266,
+      "step": 7155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617290647574887e-05,
+      "loss": 0.6429,
+      "step": 7156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616564492306515e-05,
+      "loss": 0.5368,
+      "step": 7157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761583824137367e-05,
+      "loss": 0.5317,
+      "step": 7158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7615111894785467e-05,
+      "loss": 0.569,
+      "step": 7159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614385452551036e-05,
+      "loss": 0.5559,
+      "step": 7160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7613658914679494e-05,
+      "loss": 0.5996,
+      "step": 7161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761293228117997e-05,
+      "loss": 0.6113,
+      "step": 7162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761220555206159e-05,
+      "loss": 0.4749,
+      "step": 7163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611478727333486e-05,
+      "loss": 0.6167,
+      "step": 7164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761075180700478e-05,
+      "loss": 0.5304,
+      "step": 7165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76100247910846e-05,
+      "loss": 0.5236,
+      "step": 7166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760929767958209e-05,
+      "loss": 0.5275,
+      "step": 7167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760857047250637e-05,
+      "loss": 0.5238,
+      "step": 7168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760784316986658e-05,
+      "loss": 0.5503,
+      "step": 7169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607115771671857e-05,
+      "loss": 0.6037,
+      "step": 7170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760638827793133e-05,
+      "loss": 0.5309,
+      "step": 7171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760566068865414e-05,
+      "loss": 0.5956,
+      "step": 7172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604933003849428e-05,
+      "loss": 0.5463,
+      "step": 7173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760420522352633e-05,
+      "loss": 0.5569,
+      "step": 7174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7603477347693993e-05,
+      "loss": 0.5104,
+      "step": 7175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760274937636155e-05,
+      "loss": 0.54,
+      "step": 7176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760202130953815e-05,
+      "loss": 0.5829,
+      "step": 7177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7601293147232935e-05,
+      "loss": 0.563,
+      "step": 7178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600564889455055e-05,
+      "loss": 0.5817,
+      "step": 7179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599836536213654e-05,
+      "loss": 0.5305,
+      "step": 7180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759910808751788e-05,
+      "loss": 0.5428,
+      "step": 7181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598379543376884e-05,
+      "loss": 0.5971,
+      "step": 7182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597650903799814e-05,
+      "loss": 0.5669,
+      "step": 7183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7596922168795822e-05,
+      "loss": 0.556,
+      "step": 7184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759619333837406e-05,
+      "loss": 0.5059,
+      "step": 7185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759546441254369e-05,
+      "loss": 0.5153,
+      "step": 7186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759473539131386e-05,
+      "loss": 0.5454,
+      "step": 7187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594006274693724e-05,
+      "loss": 0.6485,
+      "step": 7188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759327706269245e-05,
+      "loss": 0.5434,
+      "step": 7189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592547755319188e-05,
+      "loss": 0.4822,
+      "step": 7190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75918183525831e-05,
+      "loss": 0.5665,
+      "step": 7191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7591088854493348e-05,
+      "loss": 0.5803,
+      "step": 7192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590359261059096e-05,
+      "loss": 0.5938,
+      "step": 7193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589629572289508e-05,
+      "loss": 0.5494,
+      "step": 7194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588899788193743e-05,
+      "loss": 0.6171,
+      "step": 7195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588169908780976e-05,
+      "loss": 0.5624,
+      "step": 7196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587439934060372e-05,
+      "loss": 0.5857,
+      "step": 7197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758670986404109e-05,
+      "loss": 0.5096,
+      "step": 7198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758597969873231e-05,
+      "loss": 0.5113,
+      "step": 7199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585249438143208e-05,
+      "loss": 0.5572,
+      "step": 7200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7584519082282938e-05,
+      "loss": 0.5698,
+      "step": 7201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583788631160688e-05,
+      "loss": 0.5492,
+      "step": 7202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583058084785626e-05,
+      "loss": 0.5871,
+      "step": 7203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582327443166933e-05,
+      "loss": 0.565,
+      "step": 7204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758159670631378e-05,
+      "loss": 0.6087,
+      "step": 7205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758086587423535e-05,
+      "loss": 0.5309,
+      "step": 7206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580134946940816e-05,
+      "loss": 0.5477,
+      "step": 7207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579403924439366e-05,
+      "loss": 0.6357,
+      "step": 7208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578672806740174e-05,
+      "loss": 0.5619,
+      "step": 7209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577941593852433e-05,
+      "loss": 0.5698,
+      "step": 7210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577210285785317e-05,
+      "loss": 0.5657,
+      "step": 7211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757647888254802e-05,
+      "loss": 0.4971,
+      "step": 7212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757574738414972e-05,
+      "loss": 0.6328,
+      "step": 7213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757501579059961e-05,
+      "loss": 0.5897,
+      "step": 7214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574284101906874e-05,
+      "loss": 0.5654,
+      "step": 7215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573552318080712e-05,
+      "loss": 0.561,
+      "step": 7216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572820439130303e-05,
+      "loss": 0.5558,
+      "step": 7217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572088465064847e-05,
+      "loss": 0.5792,
+      "step": 7218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571356395893535e-05,
+      "loss": 0.5089,
+      "step": 7219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570624231625567e-05,
+      "loss": 0.5698,
+      "step": 7220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756989197227013e-05,
+      "loss": 0.5466,
+      "step": 7221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7569159617836425e-05,
+      "loss": 0.617,
+      "step": 7222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568427168333656e-05,
+      "loss": 0.5292,
+      "step": 7223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567694623771015e-05,
+      "loss": 0.6326,
+      "step": 7224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566961984157705e-05,
+      "loss": 0.6056,
+      "step": 7225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566229249502927e-05,
+      "loss": 0.5293,
+      "step": 7226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756549641981589e-05,
+      "loss": 0.5252,
+      "step": 7227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564763495105795e-05,
+      "loss": 0.4457,
+      "step": 7228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564030475381845e-05,
+      "loss": 0.5702,
+      "step": 7229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563297360653246e-05,
+      "loss": 0.5936,
+      "step": 7230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756256415092921e-05,
+      "loss": 0.5178,
+      "step": 7231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561830846218946e-05,
+      "loss": 0.5577,
+      "step": 7232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561097446531662e-05,
+      "loss": 0.5935,
+      "step": 7233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560363951876572e-05,
+      "loss": 0.5577,
+      "step": 7234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755963036226289e-05,
+      "loss": 0.5691,
+      "step": 7235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755889667769982e-05,
+      "loss": 0.5107,
+      "step": 7236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755816289819659e-05,
+      "loss": 0.5414,
+      "step": 7237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7557429023762406e-05,
+      "loss": 0.5366,
+      "step": 7238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556695054406495e-05,
+      "loss": 0.6236,
+      "step": 7239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755596099013807e-05,
+      "loss": 0.6093,
+      "step": 7240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755522683096635e-05,
+      "loss": 0.51,
+      "step": 7241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554492576900558e-05,
+      "loss": 0.5733,
+      "step": 7242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755375822794992e-05,
+      "loss": 0.5422,
+      "step": 7243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7553023784123647e-05,
+      "loss": 0.5508,
+      "step": 7244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755228924543098e-05,
+      "loss": 0.5618,
+      "step": 7245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551554611881135e-05,
+      "loss": 0.5769,
+      "step": 7246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755081988348334e-05,
+      "loss": 0.5656,
+      "step": 7247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550085060246823e-05,
+      "loss": 0.5064,
+      "step": 7248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549350142180818e-05,
+      "loss": 0.5645,
+      "step": 7249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754861512929455e-05,
+      "loss": 0.5382,
+      "step": 7250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547880021597257e-05,
+      "loss": 0.4935,
+      "step": 7251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547144819098164e-05,
+      "loss": 0.5928,
+      "step": 7252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546409521806512e-05,
+      "loss": 0.5337,
+      "step": 7253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545674129731535e-05,
+      "loss": 0.5322,
+      "step": 7254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544938642882465e-05,
+      "loss": 0.5763,
+      "step": 7255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544203061268545e-05,
+      "loss": 0.5758,
+      "step": 7256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754346738489901e-05,
+      "loss": 0.5222,
+      "step": 7257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542731613783105e-05,
+      "loss": 0.5354,
+      "step": 7258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541995747930064e-05,
+      "loss": 0.6283,
+      "step": 7259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541259787349138e-05,
+      "loss": 0.5567,
+      "step": 7260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7540523732049565e-05,
+      "loss": 0.6224,
+      "step": 7261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539787582040593e-05,
+      "loss": 0.5532,
+      "step": 7262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539051337331466e-05,
+      "loss": 0.5593,
+      "step": 7263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538314997931433e-05,
+      "loss": 0.605,
+      "step": 7264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753757856384974e-05,
+      "loss": 0.5169,
+      "step": 7265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7536842035095637e-05,
+      "loss": 0.5314,
+      "step": 7266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753610541167838e-05,
+      "loss": 0.5988,
+      "step": 7267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7535368693607214e-05,
+      "loss": 0.5771,
+      "step": 7268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534631880891397e-05,
+      "loss": 0.5647,
+      "step": 7269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533894973540176e-05,
+      "loss": 0.5629,
+      "step": 7270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753315797156282e-05,
+      "loss": 0.5447,
+      "step": 7271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753242087496857e-05,
+      "loss": 0.5781,
+      "step": 7272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531683683766697e-05,
+      "loss": 0.6097,
+      "step": 7273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530946397966454e-05,
+      "loss": 0.5685,
+      "step": 7274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530209017577108e-05,
+      "loss": 0.5714,
+      "step": 7275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529471542607906e-05,
+      "loss": 0.562,
+      "step": 7276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7528733973068127e-05,
+      "loss": 0.5608,
+      "step": 7277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527996308967022e-05,
+      "loss": 0.5814,
+      "step": 7278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527258550313863e-05,
+      "loss": 0.5748,
+      "step": 7279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526520697117918e-05,
+      "loss": 0.5405,
+      "step": 7280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525782749388448e-05,
+      "loss": 0.6057,
+      "step": 7281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752504470713473e-05,
+      "loss": 0.5706,
+      "step": 7282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524306570366024e-05,
+      "loss": 0.6319,
+      "step": 7283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523568339091607e-05,
+      "loss": 0.5336,
+      "step": 7284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752283001332075e-05,
+      "loss": 0.6121,
+      "step": 7285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752209159306273e-05,
+      "loss": 0.6764,
+      "step": 7286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521353078326817e-05,
+      "loss": 0.6163,
+      "step": 7287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520614469122287e-05,
+      "loss": 0.4951,
+      "step": 7288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519875765458416e-05,
+      "loss": 0.5345,
+      "step": 7289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519136967344485e-05,
+      "loss": 0.6213,
+      "step": 7290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.5506,
+      "step": 7291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517659087803562e-05,
+      "loss": 0.634,
+      "step": 7292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516920006395132e-05,
+      "loss": 0.5813,
+      "step": 7293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516180830573764e-05,
+      "loss": 0.5263,
+      "step": 7294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515441560348745e-05,
+      "loss": 0.5327,
+      "step": 7295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7514702195729357e-05,
+      "loss": 0.5535,
+      "step": 7296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751396273672489e-05,
+      "loss": 0.5477,
+      "step": 7297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513223183344635e-05,
+      "loss": 0.573,
+      "step": 7298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7512483535597868e-05,
+      "loss": 0.5284,
+      "step": 7299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751174379349389e-05,
+      "loss": 0.5385,
+      "step": 7300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511003957041992e-05,
+      "loss": 0.5666,
+      "step": 7301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751026402625146e-05,
+      "loss": 0.4984,
+      "step": 7302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509524001131593e-05,
+      "loss": 0.5491,
+      "step": 7303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508783881691683e-05,
+      "loss": 0.5698,
+      "step": 7304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508043667941027e-05,
+      "loss": 0.5161,
+      "step": 7305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507303359888923e-05,
+      "loss": 0.5403,
+      "step": 7306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506562957544667e-05,
+      "loss": 0.5886,
+      "step": 7307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750582246091756e-05,
+      "loss": 0.603,
+      "step": 7308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505081870016906e-05,
+      "loss": 0.6371,
+      "step": 7309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504341184851998e-05,
+      "loss": 0.5518,
+      "step": 7310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503600405432147e-05,
+      "loss": 0.5652,
+      "step": 7311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502859531766657e-05,
+      "loss": 0.5201,
+      "step": 7312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502118563864828e-05,
+      "loss": 0.5746,
+      "step": 7313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750137750173597e-05,
+      "loss": 0.5693,
+      "step": 7314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500636345389392e-05,
+      "loss": 0.5344,
+      "step": 7315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74998950948344e-05,
+      "loss": 0.5479,
+      "step": 7316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499153750080304e-05,
+      "loss": 0.5904,
+      "step": 7317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498412311136423e-05,
+      "loss": 0.5911,
+      "step": 7318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497670778012056e-05,
+      "loss": 0.507,
+      "step": 7319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496929150716527e-05,
+      "loss": 0.6076,
+      "step": 7320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749618742925915e-05,
+      "loss": 0.576,
+      "step": 7321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495445613649236e-05,
+      "loss": 0.5236,
+      "step": 7322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494703703896104e-05,
+      "loss": 0.6047,
+      "step": 7323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493961700009077e-05,
+      "loss": 0.6221,
+      "step": 7324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493219601997472e-05,
+      "loss": 0.5862,
+      "step": 7325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492477409870606e-05,
+      "loss": 0.584,
+      "step": 7326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491735123637808e-05,
+      "loss": 0.5892,
+      "step": 7327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490992743308393e-05,
+      "loss": 0.6057,
+      "step": 7328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490250268891694e-05,
+      "loss": 0.5478,
+      "step": 7329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748950770039703e-05,
+      "loss": 0.6031,
+      "step": 7330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748876503783373e-05,
+      "loss": 0.5618,
+      "step": 7331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748802228121112e-05,
+      "loss": 0.5259,
+      "step": 7332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748727943053853e-05,
+      "loss": 0.5212,
+      "step": 7333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7486536485825296e-05,
+      "loss": 0.601,
+      "step": 7334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485793447080743e-05,
+      "loss": 0.6101,
+      "step": 7335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74850503143142e-05,
+      "loss": 0.5437,
+      "step": 7336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484307087535014e-05,
+      "loss": 0.5311,
+      "step": 7337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483563766752505e-05,
+      "loss": 0.612,
+      "step": 7338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482820351976016e-05,
+      "loss": 0.5207,
+      "step": 7339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748207684321489e-05,
+      "loss": 0.6046,
+      "step": 7340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481333240478456e-05,
+      "loss": 0.5757,
+      "step": 7341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7480589543776057e-05,
+      "loss": 0.5671,
+      "step": 7342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7479845753117036e-05,
+      "loss": 0.5161,
+      "step": 7343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747910186851073e-05,
+      "loss": 0.5808,
+      "step": 7344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7478357889966487e-05,
+      "loss": 0.563,
+      "step": 7345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477613817493655e-05,
+      "loss": 0.6673,
+      "step": 7346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476869651101572e-05,
+      "loss": 0.5589,
+      "step": 7347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476125390799584e-05,
+      "loss": 0.5525,
+      "step": 7348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475381036597045e-05,
+      "loss": 0.5424,
+      "step": 7349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74746365885033e-05,
+      "loss": 0.554,
+      "step": 7350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473892046527706e-05,
+      "loss": 0.5668,
+      "step": 7351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473147410679604e-05,
+      "loss": 0.5819,
+      "step": 7352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747240268096835e-05,
+      "loss": 0.5498,
+      "step": 7353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471657857403307e-05,
+      "loss": 0.6358,
+      "step": 7354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747091293999382e-05,
+      "loss": 0.5341,
+      "step": 7355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470167928749245e-05,
+      "loss": 0.5698,
+      "step": 7356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469422823678944e-05,
+      "loss": 0.5857,
+      "step": 7357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468677624792276e-05,
+      "loss": 0.5991,
+      "step": 7358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74679323320986e-05,
+      "loss": 0.5413,
+      "step": 7359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746718694560727e-05,
+      "loss": 0.5377,
+      "step": 7360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466441465327658e-05,
+      "loss": 0.5244,
+      "step": 7361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7465695891269122e-05,
+      "loss": 0.5945,
+      "step": 7362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464950223441026e-05,
+      "loss": 0.584,
+      "step": 7363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746420446185274e-05,
+      "loss": 0.5188,
+      "step": 7364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463458606513627e-05,
+      "loss": 0.541,
+      "step": 7365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462712657433053e-05,
+      "loss": 0.5616,
+      "step": 7366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461966614620395e-05,
+      "loss": 0.5551,
+      "step": 7367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461220478085016e-05,
+      "loss": 0.5601,
+      "step": 7368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7460474247836286e-05,
+      "loss": 0.5431,
+      "step": 7369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459727923883588e-05,
+      "loss": 0.5683,
+      "step": 7370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458981506236286e-05,
+      "loss": 0.494,
+      "step": 7371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745823499490376e-05,
+      "loss": 0.4784,
+      "step": 7372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457488389895382e-05,
+      "loss": 0.5419,
+      "step": 7373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7456741691220536e-05,
+      "loss": 0.5669,
+      "step": 7374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455994898888593e-05,
+      "loss": 0.6033,
+      "step": 7375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455248012908938e-05,
+      "loss": 0.527,
+      "step": 7376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454501033290953e-05,
+      "loss": 0.5534,
+      "step": 7377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453753960044016e-05,
+      "loss": 0.6172,
+      "step": 7378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745300679317751e-05,
+      "loss": 0.5444,
+      "step": 7379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452259532700823e-05,
+      "loss": 0.5961,
+      "step": 7380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451512178623338e-05,
+      "loss": 0.5329,
+      "step": 7381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450764730954445e-05,
+      "loss": 0.5873,
+      "step": 7382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450017189703525e-05,
+      "loss": 0.5266,
+      "step": 7383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7449269554879972e-05,
+      "loss": 0.5708,
+      "step": 7384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448521826493182e-05,
+      "loss": 0.5629,
+      "step": 7385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447774004552536e-05,
+      "loss": 0.5555,
+      "step": 7386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447026089067437e-05,
+      "loss": 0.5967,
+      "step": 7387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446278080047266e-05,
+      "loss": 0.5054,
+      "step": 7388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744552997750143e-05,
+      "loss": 0.5613,
+      "step": 7389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7444781781439316e-05,
+      "loss": 0.4519,
+      "step": 7390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744403349187033e-05,
+      "loss": 0.5382,
+      "step": 7391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443285108803864e-05,
+      "loss": 0.5788,
+      "step": 7392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442536632249322e-05,
+      "loss": 0.5473,
+      "step": 7393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74417880622161e-05,
+      "loss": 0.61,
+      "step": 7394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744103939871361e-05,
+      "loss": 0.5082,
+      "step": 7395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440290641751242e-05,
+      "loss": 0.5198,
+      "step": 7396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439541791338407e-05,
+      "loss": 0.5505,
+      "step": 7397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438792847484512e-05,
+      "loss": 0.5393,
+      "step": 7398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438043810198966e-05,
+      "loss": 0.5334,
+      "step": 7399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743729467949117e-05,
+      "loss": 0.5905,
+      "step": 7400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743654545537053e-05,
+      "loss": 0.5276,
+      "step": 7401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435796137846473e-05,
+      "loss": 0.632,
+      "step": 7402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435046726928396e-05,
+      "loss": 0.5197,
+      "step": 7403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434297222625715e-05,
+      "loss": 0.581,
+      "step": 7404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7433547624947844e-05,
+      "loss": 0.5174,
+      "step": 7405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74327979339042e-05,
+      "loss": 0.567,
+      "step": 7406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432048149504198e-05,
+      "loss": 0.5196,
+      "step": 7407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743129827175725e-05,
+      "loss": 0.5686,
+      "step": 7408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7430548300672786e-05,
+      "loss": 0.564,
+      "step": 7409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429798236260215e-05,
+      "loss": 0.608,
+      "step": 7410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429048078528967e-05,
+      "loss": 0.6184,
+      "step": 7411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428297827488454e-05,
+      "loss": 0.5652,
+      "step": 7412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427547483148105e-05,
+      "loss": 0.5478,
+      "step": 7413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426797045517344e-05,
+      "loss": 0.5135,
+      "step": 7414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426046514605595e-05,
+      "loss": 0.5388,
+      "step": 7415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742529589042229e-05,
+      "loss": 0.5896,
+      "step": 7416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424545172976852e-05,
+      "loss": 0.5423,
+      "step": 7417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423794362278706e-05,
+      "loss": 0.546,
+      "step": 7418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423043458337293e-05,
+      "loss": 0.5353,
+      "step": 7419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422292461162035e-05,
+      "loss": 0.5579,
+      "step": 7420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742154137076237e-05,
+      "loss": 0.6002,
+      "step": 7421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420790187147725e-05,
+      "loss": 0.503,
+      "step": 7422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420038910327545e-05,
+      "loss": 0.582,
+      "step": 7423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419287540311257e-05,
+      "loss": 0.4997,
+      "step": 7424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418536077108307e-05,
+      "loss": 0.5285,
+      "step": 7425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741778452072812e-05,
+      "loss": 0.6159,
+      "step": 7426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741703287118015e-05,
+      "loss": 0.5624,
+      "step": 7427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741628112847383e-05,
+      "loss": 0.5833,
+      "step": 7428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415529292618606e-05,
+      "loss": 0.5658,
+      "step": 7429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414777363623918e-05,
+      "loss": 0.5315,
+      "step": 7430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741402534149921e-05,
+      "loss": 0.6198,
+      "step": 7431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741327322625393e-05,
+      "loss": 0.5031,
+      "step": 7432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7412521017897522e-05,
+      "loss": 0.5572,
+      "step": 7433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741176871643943e-05,
+      "loss": 0.5711,
+      "step": 7434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411016321889115e-05,
+      "loss": 0.534,
+      "step": 7435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410263834256018e-05,
+      "loss": 0.5526,
+      "step": 7436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409511253549592e-05,
+      "loss": 0.5572,
+      "step": 7437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740875857977929e-05,
+      "loss": 0.5236,
+      "step": 7438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408005812954566e-05,
+      "loss": 0.5308,
+      "step": 7439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740725295308487e-05,
+      "loss": 0.5065,
+      "step": 7440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406500000179666e-05,
+      "loss": 0.5605,
+      "step": 7441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405746954248407e-05,
+      "loss": 0.5941,
+      "step": 7442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740499381530055e-05,
+      "loss": 0.5775,
+      "step": 7443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404240583345554e-05,
+      "loss": 0.5657,
+      "step": 7444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403487258392887e-05,
+      "loss": 0.5408,
+      "step": 7445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402733840452e-05,
+      "loss": 0.4985,
+      "step": 7446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401980329532367e-05,
+      "loss": 0.5228,
+      "step": 7447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401226725643442e-05,
+      "loss": 0.5339,
+      "step": 7448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400473028794695e-05,
+      "loss": 0.5444,
+      "step": 7449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399719238995595e-05,
+      "loss": 0.5556,
+      "step": 7450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398965356255605e-05,
+      "loss": 0.5398,
+      "step": 7451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398211380584194e-05,
+      "loss": 0.4993,
+      "step": 7452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397457311990837e-05,
+      "loss": 0.5841,
+      "step": 7453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7396703150485002e-05,
+      "loss": 0.6192,
+      "step": 7454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395948896076157e-05,
+      "loss": 0.5835,
+      "step": 7455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395194548773784e-05,
+      "loss": 0.5988,
+      "step": 7456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739444010858735e-05,
+      "loss": 0.5653,
+      "step": 7457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739368557552634e-05,
+      "loss": 0.5673,
+      "step": 7458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7392930949600217e-05,
+      "loss": 0.5109,
+      "step": 7459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739217623081847e-05,
+      "loss": 0.5737,
+      "step": 7460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739142141919058e-05,
+      "loss": 0.4762,
+      "step": 7461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390666514726018e-05,
+      "loss": 0.488,
+      "step": 7462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.738991151743427e-05,
+      "loss": 0.5764,
+      "step": 7463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389156427324824e-05,
+      "loss": 0.5842,
+      "step": 7464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.738840124440716e-05,
+      "loss": 0.5682,
+      "step": 7465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387645968690757e-05,
+      "loss": 0.5857,
+      "step": 7466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738689060018511e-05,
+      "loss": 0.5917,
+      "step": 7467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386135138899704e-05,
+      "loss": 0.5543,
+      "step": 7468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385379584844026e-05,
+      "loss": 0.6305,
+      "step": 7469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384623938027565e-05,
+      "loss": 0.5077,
+      "step": 7470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383868198459817e-05,
+      "loss": 0.4765,
+      "step": 7471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738311236615027e-05,
+      "loss": 0.5638,
+      "step": 7472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382356441108417e-05,
+      "loss": 0.6006,
+      "step": 7473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381600423343753e-05,
+      "loss": 0.5322,
+      "step": 7474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380844312865773e-05,
+      "loss": 0.5521,
+      "step": 7475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738008810968398e-05,
+      "loss": 0.6124,
+      "step": 7476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379331813807866e-05,
+      "loss": 0.5856,
+      "step": 7477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737857542524693e-05,
+      "loss": 0.582,
+      "step": 7478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377818944010672e-05,
+      "loss": 0.5405,
+      "step": 7479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377062370108596e-05,
+      "loss": 0.557,
+      "step": 7480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376305703550206e-05,
+      "loss": 0.577,
+      "step": 7481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375548944344996e-05,
+      "loss": 0.5729,
+      "step": 7482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374792092502483e-05,
+      "loss": 0.5945,
+      "step": 7483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737403514803217e-05,
+      "loss": 0.5724,
+      "step": 7484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737327811094356e-05,
+      "loss": 0.5477,
+      "step": 7485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7372520981246166e-05,
+      "loss": 0.5574,
+      "step": 7486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371763758949494e-05,
+      "loss": 0.5818,
+      "step": 7487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371006444063057e-05,
+      "loss": 0.5075,
+      "step": 7488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370249036596366e-05,
+      "loss": 0.5034,
+      "step": 7489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7369491536558932e-05,
+      "loss": 0.4937,
+      "step": 7490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368733943960278e-05,
+      "loss": 0.5793,
+      "step": 7491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367976258809905e-05,
+      "loss": 0.5665,
+      "step": 7492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736721848111734e-05,
+      "loss": 0.5605,
+      "step": 7493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366460610892102e-05,
+      "loss": 0.5056,
+      "step": 7494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365702648143703e-05,
+      "loss": 0.5731,
+      "step": 7495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364944592881668e-05,
+      "loss": 0.5281,
+      "step": 7496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364186445115513e-05,
+      "loss": 0.5698,
+      "step": 7497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363428204854766e-05,
+      "loss": 0.5817,
+      "step": 7498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7362669872108952e-05,
+      "loss": 0.4834,
+      "step": 7499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361911446887587e-05,
+      "loss": 0.621,
+      "step": 7500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361152929200203e-05,
+      "loss": 0.4908,
+      "step": 7501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360394319056326e-05,
+      "loss": 0.5456,
+      "step": 7502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359635616465485e-05,
+      "loss": 0.576,
+      "step": 7503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358876821437206e-05,
+      "loss": 0.57,
+      "step": 7504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358117933981024e-05,
+      "loss": 0.4692,
+      "step": 7505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735735895410647e-05,
+      "loss": 0.5165,
+      "step": 7506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356599881823074e-05,
+      "loss": 0.6013,
+      "step": 7507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735584071714037e-05,
+      "loss": 0.5734,
+      "step": 7508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7355081460067897e-05,
+      "loss": 0.4832,
+      "step": 7509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735432211061519e-05,
+      "loss": 0.5335,
+      "step": 7510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353562668791783e-05,
+      "loss": 0.5542,
+      "step": 7511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352803134607216e-05,
+      "loss": 0.6027,
+      "step": 7512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735204350807103e-05,
+      "loss": 0.5344,
+      "step": 7513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735128378919277e-05,
+      "loss": 0.6442,
+      "step": 7514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350523977981966e-05,
+      "loss": 0.6401,
+      "step": 7515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349764074448178e-05,
+      "loss": 0.5106,
+      "step": 7516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349004078600938e-05,
+      "loss": 0.5472,
+      "step": 7517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348243990449793e-05,
+      "loss": 0.602,
+      "step": 7518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347483810004295e-05,
+      "loss": 0.5566,
+      "step": 7519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734672353727399e-05,
+      "loss": 0.5857,
+      "step": 7520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734596317226842e-05,
+      "loss": 0.5679,
+      "step": 7521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345202714997147e-05,
+      "loss": 0.513,
+      "step": 7522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344442165469714e-05,
+      "loss": 0.5523,
+      "step": 7523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343681523695676e-05,
+      "loss": 0.5693,
+      "step": 7524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342920789684586e-05,
+      "loss": 0.5573,
+      "step": 7525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342159963446e-05,
+      "loss": 0.5412,
+      "step": 7526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341399044989478e-05,
+      "loss": 0.5476,
+      "step": 7527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340638034324565e-05,
+      "loss": 0.5772,
+      "step": 7528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339876931460832e-05,
+      "loss": 0.5626,
+      "step": 7529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339115736407835e-05,
+      "loss": 0.555,
+      "step": 7530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338354449175132e-05,
+      "loss": 0.5549,
+      "step": 7531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337593069772285e-05,
+      "loss": 0.5024,
+      "step": 7532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336831598208855e-05,
+      "loss": 0.5566,
+      "step": 7533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336070034494413e-05,
+      "loss": 0.5306,
+      "step": 7534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335308378638522e-05,
+      "loss": 0.5395,
+      "step": 7535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7334546630650745e-05,
+      "loss": 0.5653,
+      "step": 7536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333784790540654e-05,
+      "loss": 0.5435,
+      "step": 7537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333022858317813e-05,
+      "loss": 0.4851,
+      "step": 7538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332260833991795e-05,
+      "loss": 0.5194,
+      "step": 7539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331498717572167e-05,
+      "loss": 0.5835,
+      "step": 7540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330736509068507e-05,
+      "loss": 0.4958,
+      "step": 7541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329974208490384e-05,
+      "loss": 0.5683,
+      "step": 7542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329211815847376e-05,
+      "loss": 0.5469,
+      "step": 7543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732844933114906e-05,
+      "loss": 0.5169,
+      "step": 7544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327686754405007e-05,
+      "loss": 0.4953,
+      "step": 7545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73269240856248e-05,
+      "loss": 0.6692,
+      "step": 7546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326161324818013e-05,
+      "loss": 0.5573,
+      "step": 7547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325398471994232e-05,
+      "loss": 0.5169,
+      "step": 7548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324635527163033e-05,
+      "loss": 0.6139,
+      "step": 7549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323872490334002e-05,
+      "loss": 0.5617,
+      "step": 7550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323109361516726e-05,
+      "loss": 0.5281,
+      "step": 7551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322346140720786e-05,
+      "loss": 0.5267,
+      "step": 7552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321582827955764e-05,
+      "loss": 0.5527,
+      "step": 7553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320819423231254e-05,
+      "loss": 0.5253,
+      "step": 7554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320055926556843e-05,
+      "loss": 0.5552,
+      "step": 7555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731929233794212e-05,
+      "loss": 0.4562,
+      "step": 7556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318528657396674e-05,
+      "loss": 0.5599,
+      "step": 7557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317764884930096e-05,
+      "loss": 0.565,
+      "step": 7558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317001020551984e-05,
+      "loss": 0.5987,
+      "step": 7559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731623706427193e-05,
+      "loss": 0.5441,
+      "step": 7560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7315473016099528e-05,
+      "loss": 0.5529,
+      "step": 7561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7314708876044375e-05,
+      "loss": 0.5344,
+      "step": 7562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313944644116065e-05,
+      "loss": 0.6759,
+      "step": 7563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731318032032421e-05,
+      "loss": 0.5273,
+      "step": 7564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731241590467839e-05,
+      "loss": 0.5315,
+      "step": 7565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731165139718822e-05,
+      "loss": 0.5509,
+      "step": 7566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310886797863304e-05,
+      "loss": 0.5595,
+      "step": 7567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310122106713235e-05,
+      "loss": 0.6289,
+      "step": 7568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309357323747624e-05,
+      "loss": 0.5853,
+      "step": 7569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730859244897608e-05,
+      "loss": 0.5152,
+      "step": 7570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73078274824082e-05,
+      "loss": 0.5453,
+      "step": 7571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73070624240536e-05,
+      "loss": 0.5352,
+      "step": 7572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306297273921882e-05,
+      "loss": 0.4955,
+      "step": 7573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730553203202267e-05,
+      "loss": 0.5628,
+      "step": 7574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730476669836556e-05,
+      "loss": 0.5218,
+      "step": 7575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304001272960174e-05,
+      "loss": 0.4914,
+      "step": 7576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730323575581612e-05,
+      "loss": 0.5852,
+      "step": 7577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730247014694302e-05,
+      "loss": 0.5275,
+      "step": 7578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730170444635048e-05,
+      "loss": 0.4923,
+      "step": 7579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300938654048128e-05,
+      "loss": 0.5844,
+      "step": 7580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300172770045578e-05,
+      "loss": 0.5519,
+      "step": 7581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299406794352447e-05,
+      "loss": 0.5147,
+      "step": 7582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298640726978357e-05,
+      "loss": 0.5273,
+      "step": 7583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297874567932933e-05,
+      "loss": 0.5454,
+      "step": 7584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297108317225798e-05,
+      "loss": 0.5982,
+      "step": 7585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7296341974866568e-05,
+      "loss": 0.5526,
+      "step": 7586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295575540864878e-05,
+      "loss": 0.5325,
+      "step": 7587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294809015230347e-05,
+      "loss": 0.4965,
+      "step": 7588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294042397972612e-05,
+      "loss": 0.5414,
+      "step": 7589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729327568910129e-05,
+      "loss": 0.5616,
+      "step": 7590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292508888626022e-05,
+      "loss": 0.5327,
+      "step": 7591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729174199655643e-05,
+      "loss": 0.5257,
+      "step": 7592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729097501290215e-05,
+      "loss": 0.5462,
+      "step": 7593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729020793767282e-05,
+      "loss": 0.5571,
+      "step": 7594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289440770878065e-05,
+      "loss": 0.5822,
+      "step": 7595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728867351252753e-05,
+      "loss": 0.5514,
+      "step": 7596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728790616263084e-05,
+      "loss": 0.5813,
+      "step": 7597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7287138721197648e-05,
+      "loss": 0.5648,
+      "step": 7598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286371188237584e-05,
+      "loss": 0.5303,
+      "step": 7599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728560356376029e-05,
+      "loss": 0.568,
+      "step": 7600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284835847775406e-05,
+      "loss": 0.5657,
+      "step": 7601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284068040292577e-05,
+      "loss": 0.5452,
+      "step": 7602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728330014132144e-05,
+      "loss": 0.537,
+      "step": 7603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282532150871652e-05,
+      "loss": 0.5393,
+      "step": 7604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281764068952852e-05,
+      "loss": 0.5417,
+      "step": 7605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280995895574683e-05,
+      "loss": 0.5016,
+      "step": 7606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280227630746802e-05,
+      "loss": 0.5268,
+      "step": 7607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279459274478852e-05,
+      "loss": 0.5621,
+      "step": 7608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278690826780488e-05,
+      "loss": 0.5549,
+      "step": 7609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277922287661353e-05,
+      "loss": 0.5443,
+      "step": 7610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277153657131114e-05,
+      "loss": 0.5539,
+      "step": 7611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276384935199417e-05,
+      "loss": 0.5199,
+      "step": 7612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275616121875913e-05,
+      "loss": 0.5858,
+      "step": 7613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274847217170267e-05,
+      "loss": 0.543,
+      "step": 7614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727407822109213e-05,
+      "loss": 0.5814,
+      "step": 7615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273309133651164e-05,
+      "loss": 0.6076,
+      "step": 7616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7272539954857027e-05,
+      "loss": 0.5072,
+      "step": 7617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271770684719382e-05,
+      "loss": 0.5071,
+      "step": 7618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271001323247892e-05,
+      "loss": 0.5116,
+      "step": 7619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270231870452217e-05,
+      "loss": 0.5528,
+      "step": 7620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269462326342022e-05,
+      "loss": 0.606,
+      "step": 7621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726869269092697e-05,
+      "loss": 0.5099,
+      "step": 7622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267922964216735e-05,
+      "loss": 0.5854,
+      "step": 7623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267153146220978e-05,
+      "loss": 0.5474,
+      "step": 7624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726638323694937e-05,
+      "loss": 0.5747,
+      "step": 7625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265613236411583e-05,
+      "loss": 0.5586,
+      "step": 7626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264843144617288e-05,
+      "loss": 0.5495,
+      "step": 7627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264072961576155e-05,
+      "loss": 0.504,
+      "step": 7628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726330268729786e-05,
+      "loss": 0.5306,
+      "step": 7629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7262532321792073e-05,
+      "loss": 0.6074,
+      "step": 7630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261761865068476e-05,
+      "loss": 0.5667,
+      "step": 7631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7260991317136743e-05,
+      "loss": 0.5042,
+      "step": 7632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7260220678006553e-05,
+      "loss": 0.5952,
+      "step": 7633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259449947687584e-05,
+      "loss": 0.5475,
+      "step": 7634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7258679126189515e-05,
+      "loss": 0.5673,
+      "step": 7635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257908213522032e-05,
+      "loss": 0.596,
+      "step": 7636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257137209694816e-05,
+      "loss": 0.5873,
+      "step": 7637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256366114717548e-05,
+      "loss": 0.6232,
+      "step": 7638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255594928599918e-05,
+      "loss": 0.515,
+      "step": 7639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725482365135161e-05,
+      "loss": 0.5252,
+      "step": 7640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254052282982307e-05,
+      "loss": 0.5252,
+      "step": 7641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7253280823501703e-05,
+      "loss": 0.551,
+      "step": 7642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7252509272919483e-05,
+      "loss": 0.5732,
+      "step": 7643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725173763124534e-05,
+      "loss": 0.5349,
+      "step": 7644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725096589848897e-05,
+      "loss": 0.5809,
+      "step": 7645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250194074660062e-05,
+      "loss": 0.5087,
+      "step": 7646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249422159768308e-05,
+      "loss": 0.6248,
+      "step": 7647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724865015382341e-05,
+      "loss": 0.5704,
+      "step": 7648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247878056835055e-05,
+      "loss": 0.5496,
+      "step": 7649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724710586881295e-05,
+      "loss": 0.4586,
+      "step": 7650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246333589766786e-05,
+      "loss": 0.5698,
+      "step": 7651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7245561219706265e-05,
+      "loss": 0.5403,
+      "step": 7652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7244788758641094e-05,
+      "loss": 0.4829,
+      "step": 7653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7244016206580965e-05,
+      "loss": 0.5539,
+      "step": 7654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724324356353559e-05,
+      "loss": 0.5764,
+      "step": 7655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242470829514674e-05,
+      "loss": 0.5809,
+      "step": 7656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7241698004527912e-05,
+      "loss": 0.4643,
+      "step": 7657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7240925088585018e-05,
+      "loss": 0.567,
+      "step": 7658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72401520816957e-05,
+      "loss": 0.6229,
+      "step": 7659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723937898386967e-05,
+      "loss": 0.5965,
+      "step": 7660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7238605795116633e-05,
+      "loss": 0.5485,
+      "step": 7661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72378325154463e-05,
+      "loss": 0.6017,
+      "step": 7662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7237059144868386e-05,
+      "loss": 0.5395,
+      "step": 7663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236285683392608e-05,
+      "loss": 0.5143,
+      "step": 7664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235512131028673e-05,
+      "loss": 0.5319,
+      "step": 7665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234738487786298e-05,
+      "loss": 0.5534,
+      "step": 7666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233964753675205e-05,
+      "loss": 0.5349,
+      "step": 7667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723319092870511e-05,
+      "loss": 0.5551,
+      "step": 7668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723241701288573e-05,
+      "loss": 0.5177,
+      "step": 7669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7231643006226787e-05,
+      "loss": 0.5483,
+      "step": 7670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230868908738007e-05,
+      "loss": 0.5736,
+      "step": 7671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230094720429104e-05,
+      "loss": 0.5332,
+      "step": 7672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229320441309806e-05,
+      "loss": 0.5376,
+      "step": 7673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722854607138984e-05,
+      "loss": 0.4849,
+      "step": 7674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227771610678932e-05,
+      "loss": 0.4972,
+      "step": 7675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226997059186805e-05,
+      "loss": 0.5491,
+      "step": 7676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226222416923194e-05,
+      "loss": 0.5402,
+      "step": 7677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722544768389782e-05,
+      "loss": 0.6542,
+      "step": 7678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722467286012042e-05,
+      "loss": 0.5831,
+      "step": 7679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223897945600726e-05,
+      "loss": 0.6018,
+      "step": 7680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223122940348468e-05,
+      "loss": 0.5469,
+      "step": 7681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222347844373387e-05,
+      "loss": 0.562,
+      "step": 7682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221572657685205e-05,
+      "loss": 0.5295,
+      "step": 7683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220797380293664e-05,
+      "loss": 0.4999,
+      "step": 7684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722002201220851e-05,
+      "loss": 0.5079,
+      "step": 7685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219246553439475e-05,
+      "loss": 0.5591,
+      "step": 7686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7218471003996297e-05,
+      "loss": 0.5217,
+      "step": 7687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217695363888718e-05,
+      "loss": 0.5565,
+      "step": 7688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216919633126482e-05,
+      "loss": 0.549,
+      "step": 7689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721614381171933e-05,
+      "loss": 0.5065,
+      "step": 7690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215367899677008e-05,
+      "loss": 0.573,
+      "step": 7691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721459189700926e-05,
+      "loss": 0.5347,
+      "step": 7692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721381580372584e-05,
+      "loss": 0.5508,
+      "step": 7693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213039619836484e-05,
+      "loss": 0.5709,
+      "step": 7694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212263345350945e-05,
+      "loss": 0.5552,
+      "step": 7695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211486980278976e-05,
+      "loss": 0.5557,
+      "step": 7696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721071052463033e-05,
+      "loss": 0.5234,
+      "step": 7697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209933978414753e-05,
+      "loss": 0.5915,
+      "step": 7698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209157341642e-05,
+      "loss": 0.6896,
+      "step": 7699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720838061432183e-05,
+      "loss": 0.6087,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207603796463995e-05,
+      "loss": 0.5193,
+      "step": 7701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206826888078253e-05,
+      "loss": 0.5824,
+      "step": 7702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720604988917436e-05,
+      "loss": 0.4821,
+      "step": 7703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7205272799762078e-05,
+      "loss": 0.5891,
+      "step": 7704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204495619851166e-05,
+      "loss": 0.56,
+      "step": 7705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203718349451386e-05,
+      "loss": 0.5447,
+      "step": 7706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7202940988572495e-05,
+      "loss": 0.5203,
+      "step": 7707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720216353722427e-05,
+      "loss": 0.6464,
+      "step": 7708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201385995416464e-05,
+      "loss": 0.5435,
+      "step": 7709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200608363158845e-05,
+      "loss": 0.5983,
+      "step": 7710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199830640461184e-05,
+      "loss": 0.5678,
+      "step": 7711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199052827333246e-05,
+      "loss": 0.517,
+      "step": 7712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198274923784798e-05,
+      "loss": 0.6356,
+      "step": 7713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719749692982562e-05,
+      "loss": 0.6078,
+      "step": 7714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196718845465472e-05,
+      "loss": 0.5827,
+      "step": 7715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195940670714137e-05,
+      "loss": 0.6414,
+      "step": 7716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719516240558138e-05,
+      "loss": 0.5877,
+      "step": 7717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194384050076983e-05,
+      "loss": 0.5496,
+      "step": 7718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719360560421072e-05,
+      "loss": 0.522,
+      "step": 7719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192827067992364e-05,
+      "loss": 0.5272,
+      "step": 7720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71920484414317e-05,
+      "loss": 0.5592,
+      "step": 7721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191269724538503e-05,
+      "loss": 0.5462,
+      "step": 7722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190490917322557e-05,
+      "loss": 0.53,
+      "step": 7723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189712019793644e-05,
+      "loss": 0.4608,
+      "step": 7724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188933031961542e-05,
+      "loss": 0.4891,
+      "step": 7725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718815395383604e-05,
+      "loss": 0.6219,
+      "step": 7726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187374785426923e-05,
+      "loss": 0.6039,
+      "step": 7727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718659552674397e-05,
+      "loss": 0.5258,
+      "step": 7728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718581617779698e-05,
+      "loss": 0.5971,
+      "step": 7729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185036738595736e-05,
+      "loss": 0.5295,
+      "step": 7730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718425720915003e-05,
+      "loss": 0.5575,
+      "step": 7731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718347758946965e-05,
+      "loss": 0.5633,
+      "step": 7732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718269787956439e-05,
+      "loss": 0.5955,
+      "step": 7733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181918079444038e-05,
+      "loss": 0.542,
+      "step": 7734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181138189118396e-05,
+      "loss": 0.6039,
+      "step": 7735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180358208597253e-05,
+      "loss": 0.5634,
+      "step": 7736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179578137890415e-05,
+      "loss": 0.5174,
+      "step": 7737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7178797977007667e-05,
+      "loss": 0.5314,
+      "step": 7738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7178017725958817e-05,
+      "loss": 0.4838,
+      "step": 7739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717723738475366e-05,
+      "loss": 0.5406,
+      "step": 7740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176456953402003e-05,
+      "loss": 0.5722,
+      "step": 7741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7175676431913646e-05,
+      "loss": 0.51,
+      "step": 7742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717489582029839e-05,
+      "loss": 0.5649,
+      "step": 7743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717411511856604e-05,
+      "loss": 0.6093,
+      "step": 7744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7173334326726398e-05,
+      "loss": 0.6296,
+      "step": 7745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717255344478928e-05,
+      "loss": 0.6072,
+      "step": 7746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171772472764488e-05,
+      "loss": 0.5609,
+      "step": 7747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717099141066183e-05,
+      "loss": 0.5523,
+      "step": 7748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170210258491123e-05,
+      "loss": 0.5743,
+      "step": 7749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716942901626217e-05,
+      "loss": 0.4723,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7168647683984786e-05,
+      "loss": 0.5977,
+      "step": 7751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7167866261668788e-05,
+      "loss": 0.5642,
+      "step": 7752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716708474932399e-05,
+      "loss": 0.5429,
+      "step": 7753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7166303146960203e-05,
+      "loss": 0.5471,
+      "step": 7754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165521454587246e-05,
+      "loss": 0.5105,
+      "step": 7755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164739672214944e-05,
+      "loss": 0.5275,
+      "step": 7756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163957799853103e-05,
+      "loss": 0.57,
+      "step": 7757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163175837511555e-05,
+      "loss": 0.6066,
+      "step": 7758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716239378520012e-05,
+      "loss": 0.6109,
+      "step": 7759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716161164292861e-05,
+      "loss": 0.5209,
+      "step": 7760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716082941070686e-05,
+      "loss": 0.5416,
+      "step": 7761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160047088544694e-05,
+      "loss": 0.5664,
+      "step": 7762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159264676451935e-05,
+      "loss": 0.5752,
+      "step": 7763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715848217443841e-05,
+      "loss": 0.6006,
+      "step": 7764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157699582513947e-05,
+      "loss": 0.5569,
+      "step": 7765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7156916900688377e-05,
+      "loss": 0.5988,
+      "step": 7766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7156134128971533e-05,
+      "loss": 0.5453,
+      "step": 7767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715535126737324e-05,
+      "loss": 0.5455,
+      "step": 7768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154568315903336e-05,
+      "loss": 0.5239,
+      "step": 7769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153785274571653e-05,
+      "loss": 0.5101,
+      "step": 7770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153002143388023e-05,
+      "loss": 0.5106,
+      "step": 7771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152218922362292e-05,
+      "loss": 0.5734,
+      "step": 7772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151435611504285e-05,
+      "loss": 0.5905,
+      "step": 7773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7150652210823848e-05,
+      "loss": 0.564,
+      "step": 7774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714986872033082e-05,
+      "loss": 0.5739,
+      "step": 7775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149085140035038e-05,
+      "loss": 0.6052,
+      "step": 7776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714830146994635e-05,
+      "loss": 0.5758,
+      "step": 7777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147517710074592e-05,
+      "loss": 0.5434,
+      "step": 7778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146733860429614e-05,
+      "loss": 0.5085,
+      "step": 7779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145949921021253e-05,
+      "loss": 0.5476,
+      "step": 7780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145165891859365e-05,
+      "loss": 0.5641,
+      "step": 7781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144381772953793e-05,
+      "loss": 0.5533,
+      "step": 7782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143597564314384e-05,
+      "loss": 0.5967,
+      "step": 7783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714281326595099e-05,
+      "loss": 0.5291,
+      "step": 7784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142028877873464e-05,
+      "loss": 0.5343,
+      "step": 7785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141244400091654e-05,
+      "loss": 0.532,
+      "step": 7786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7140459832615414e-05,
+      "loss": 0.6036,
+      "step": 7787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139675175454597e-05,
+      "loss": 0.5182,
+      "step": 7788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138890428619063e-05,
+      "loss": 0.5088,
+      "step": 7789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138105592118664e-05,
+      "loss": 0.5713,
+      "step": 7790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137320665963257e-05,
+      "loss": 0.5553,
+      "step": 7791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136535650162704e-05,
+      "loss": 0.584,
+      "step": 7792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135750544726864e-05,
+      "loss": 0.556,
+      "step": 7793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134965349665597e-05,
+      "loss": 0.5984,
+      "step": 7794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134180064988767e-05,
+      "loss": 0.5091,
+      "step": 7795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133394690706235e-05,
+      "loss": 0.6166,
+      "step": 7796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7132609226827868e-05,
+      "loss": 0.559,
+      "step": 7797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131823673363526e-05,
+      "loss": 0.5469,
+      "step": 7798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713103803032308e-05,
+      "loss": 0.5483,
+      "step": 7799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71302522977164e-05,
+      "loss": 0.4817,
+      "step": 7800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712946647555335e-05,
+      "loss": 0.4953,
+      "step": 7801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71286805638438e-05,
+      "loss": 0.6188,
+      "step": 7802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127894562597626e-05,
+      "loss": 0.5253,
+      "step": 7803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127108471824693e-05,
+      "loss": 0.5116,
+      "step": 7804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712632229153488e-05,
+      "loss": 0.5488,
+      "step": 7805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125536021738063e-05,
+      "loss": 0.5821,
+      "step": 7806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124749662444112e-05,
+      "loss": 0.5629,
+      "step": 7807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7123963213662905e-05,
+      "loss": 0.6007,
+      "step": 7808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7123176675404326e-05,
+      "loss": 0.5273,
+      "step": 7809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122390047678244e-05,
+      "loss": 0.5708,
+      "step": 7810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121603330494547e-05,
+      "loss": 0.5103,
+      "step": 7811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712081652386311e-05,
+      "loss": 0.5693,
+      "step": 7812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120029627793822e-05,
+      "loss": 0.6187,
+      "step": 7813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711924264229656e-05,
+      "loss": 0.5547,
+      "step": 7814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118455567381214e-05,
+      "loss": 0.5513,
+      "step": 7815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117668403057666e-05,
+      "loss": 0.5173,
+      "step": 7816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116881149335805e-05,
+      "loss": 0.5432,
+      "step": 7817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711609380622552e-05,
+      "loss": 0.5815,
+      "step": 7818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115306373736695e-05,
+      "loss": 0.5507,
+      "step": 7819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114518851879225e-05,
+      "loss": 0.6026,
+      "step": 7820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113731240663002e-05,
+      "loss": 0.5014,
+      "step": 7821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112943540097912e-05,
+      "loss": 0.5096,
+      "step": 7822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112155750193855e-05,
+      "loss": 0.5523,
+      "step": 7823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111367870960723e-05,
+      "loss": 0.5683,
+      "step": 7824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711057990240841e-05,
+      "loss": 0.5938,
+      "step": 7825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710979184454682e-05,
+      "loss": 0.5676,
+      "step": 7826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109003697385844e-05,
+      "loss": 0.5771,
+      "step": 7827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710821546093538e-05,
+      "loss": 0.6165,
+      "step": 7828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107427135205336e-05,
+      "loss": 0.5063,
+      "step": 7829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710663872020561e-05,
+      "loss": 0.4943,
+      "step": 7830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71058502159461e-05,
+      "loss": 0.5488,
+      "step": 7831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105061622436716e-05,
+      "loss": 0.5531,
+      "step": 7832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710427293968736e-05,
+      "loss": 0.485,
+      "step": 7833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710348416770794e-05,
+      "loss": 0.5636,
+      "step": 7834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102695306508353e-05,
+      "loss": 0.5189,
+      "step": 7835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101906356098522e-05,
+      "loss": 0.5476,
+      "step": 7836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101117316488346e-05,
+      "loss": 0.5928,
+      "step": 7837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100328187687742e-05,
+      "loss": 0.6367,
+      "step": 7838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099538969706617e-05,
+      "loss": 0.6193,
+      "step": 7839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098749662554884e-05,
+      "loss": 0.5611,
+      "step": 7840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709796026624246e-05,
+      "loss": 0.555,
+      "step": 7841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097170780779254e-05,
+      "loss": 0.5932,
+      "step": 7842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7096381206175188e-05,
+      "loss": 0.5602,
+      "step": 7843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709559154244018e-05,
+      "loss": 0.5608,
+      "step": 7844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7094801789584137e-05,
+      "loss": 0.5375,
+      "step": 7845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709401194761699e-05,
+      "loss": 0.5833,
+      "step": 7846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093222016548657e-05,
+      "loss": 0.5805,
+      "step": 7847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7092431996389055e-05,
+      "loss": 0.5072,
+      "step": 7848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7091641887148112e-05,
+      "loss": 0.5572,
+      "step": 7849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7090851688835746e-05,
+      "loss": 0.5921,
+      "step": 7850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709006140146189e-05,
+      "loss": 0.5784,
+      "step": 7851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7089271025036468e-05,
+      "loss": 0.6017,
+      "step": 7852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088480559569402e-05,
+      "loss": 0.5838,
+      "step": 7853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087690005070622e-05,
+      "loss": 0.5247,
+      "step": 7854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708689936155006e-05,
+      "loss": 0.5839,
+      "step": 7855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086108629017645e-05,
+      "loss": 0.5438,
+      "step": 7856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708531780748331e-05,
+      "loss": 0.5282,
+      "step": 7857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084526896956984e-05,
+      "loss": 0.55,
+      "step": 7858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083735897448602e-05,
+      "loss": 0.5198,
+      "step": 7859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082944808968102e-05,
+      "loss": 0.5438,
+      "step": 7860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708215363152542e-05,
+      "loss": 0.5553,
+      "step": 7861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708136236513049e-05,
+      "loss": 0.5584,
+      "step": 7862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080571009793254e-05,
+      "loss": 0.4771,
+      "step": 7863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707977956552365e-05,
+      "loss": 0.5929,
+      "step": 7864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078988032331614e-05,
+      "loss": 0.5458,
+      "step": 7865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078196410227093e-05,
+      "loss": 0.5499,
+      "step": 7866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077404699220034e-05,
+      "loss": 0.5374,
+      "step": 7867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707661289932037e-05,
+      "loss": 0.5971,
+      "step": 7868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075821010538053e-05,
+      "loss": 0.5125,
+      "step": 7869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707502903288303e-05,
+      "loss": 0.5619,
+      "step": 7870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074236966365244e-05,
+      "loss": 0.526,
+      "step": 7871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073444810994645e-05,
+      "loss": 0.5737,
+      "step": 7872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072652566781186e-05,
+      "loss": 0.5954,
+      "step": 7873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707186023373481e-05,
+      "loss": 0.6165,
+      "step": 7874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.571,
+      "step": 7875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070275301183134e-05,
+      "loss": 0.5765,
+      "step": 7876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706948270169774e-05,
+      "loss": 0.5141,
+      "step": 7877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068690013419246e-05,
+      "loss": 0.5395,
+      "step": 7878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706789723635761e-05,
+      "loss": 0.5489,
+      "step": 7879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067104370522788e-05,
+      "loss": 0.5163,
+      "step": 7880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066311415924737e-05,
+      "loss": 0.4802,
+      "step": 7881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065518372573426e-05,
+      "loss": 0.5756,
+      "step": 7882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70647252404788e-05,
+      "loss": 0.625,
+      "step": 7883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063932019650834e-05,
+      "loss": 0.5476,
+      "step": 7884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063138710099483e-05,
+      "loss": 0.5906,
+      "step": 7885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706234531183472e-05,
+      "loss": 0.5563,
+      "step": 7886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061551824866497e-05,
+      "loss": 0.5349,
+      "step": 7887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060758249204794e-05,
+      "loss": 0.5556,
+      "step": 7888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059964584859568e-05,
+      "loss": 0.4491,
+      "step": 7889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059170831840794e-05,
+      "loss": 0.509,
+      "step": 7890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705837699015844e-05,
+      "loss": 0.5813,
+      "step": 7891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057583059822472e-05,
+      "loss": 0.5335,
+      "step": 7892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056789040842867e-05,
+      "loss": 0.5003,
+      "step": 7893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055994933229596e-05,
+      "loss": 0.509,
+      "step": 7894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055200736992633e-05,
+      "loss": 0.6032,
+      "step": 7895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054406452141956e-05,
+      "loss": 0.563,
+      "step": 7896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053612078687536e-05,
+      "loss": 0.539,
+      "step": 7897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052817616639356e-05,
+      "loss": 0.6125,
+      "step": 7898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052023066007392e-05,
+      "loss": 0.5801,
+      "step": 7899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705122842680162e-05,
+      "loss": 0.5333,
+      "step": 7900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050433699032028e-05,
+      "loss": 0.5238,
+      "step": 7901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704963888270859e-05,
+      "loss": 0.5205,
+      "step": 7902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048843977841295e-05,
+      "loss": 0.5935,
+      "step": 7903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048048984440124e-05,
+      "loss": 0.5753,
+      "step": 7904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047253902515062e-05,
+      "loss": 0.5883,
+      "step": 7905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046458732076097e-05,
+      "loss": 0.5625,
+      "step": 7906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045663473133215e-05,
+      "loss": 0.5127,
+      "step": 7907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704486812569641e-05,
+      "loss": 0.6181,
+      "step": 7908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7044072689775665e-05,
+      "loss": 0.6345,
+      "step": 7909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043277165380967e-05,
+      "loss": 0.5681,
+      "step": 7910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704248155252232e-05,
+      "loss": 0.6013,
+      "step": 7911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041685851209706e-05,
+      "loss": 0.5444,
+      "step": 7912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040890061453125e-05,
+      "loss": 0.4959,
+      "step": 7913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704009418326257e-05,
+      "loss": 0.5399,
+      "step": 7914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703929821664804e-05,
+      "loss": 0.5165,
+      "step": 7915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038502161619526e-05,
+      "loss": 0.6103,
+      "step": 7916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703770601818703e-05,
+      "loss": 0.5054,
+      "step": 7917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036909786360557e-05,
+      "loss": 0.4892,
+      "step": 7918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70361134661501e-05,
+      "loss": 0.6253,
+      "step": 7919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035317057565666e-05,
+      "loss": 0.4987,
+      "step": 7920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034520560617256e-05,
+      "loss": 0.5406,
+      "step": 7921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703372397531487e-05,
+      "loss": 0.5205,
+      "step": 7922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703292730166852e-05,
+      "loss": 0.5099,
+      "step": 7923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703213053968821e-05,
+      "loss": 0.502,
+      "step": 7924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031333689383945e-05,
+      "loss": 0.5522,
+      "step": 7925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030536750765733e-05,
+      "loss": 0.6292,
+      "step": 7926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702973972384359e-05,
+      "loss": 0.4557,
+      "step": 7927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028942608627515e-05,
+      "loss": 0.57,
+      "step": 7928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028145405127538e-05,
+      "loss": 0.5951,
+      "step": 7929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7027348113353656e-05,
+      "loss": 0.5776,
+      "step": 7930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026550733315884e-05,
+      "loss": 0.6215,
+      "step": 7931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025753265024246e-05,
+      "loss": 0.5044,
+      "step": 7932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024955708488753e-05,
+      "loss": 0.5057,
+      "step": 7933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702415806371942e-05,
+      "loss": 0.4972,
+      "step": 7934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023360330726272e-05,
+      "loss": 0.5541,
+      "step": 7935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022562509519325e-05,
+      "loss": 0.5192,
+      "step": 7936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021764600108594e-05,
+      "loss": 0.5335,
+      "step": 7937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020966602504112e-05,
+      "loss": 0.5011,
+      "step": 7938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020168516715894e-05,
+      "loss": 0.5493,
+      "step": 7939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701937034275397e-05,
+      "loss": 0.5625,
+      "step": 7940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018572080628355e-05,
+      "loss": 0.6251,
+      "step": 7941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017773730349084e-05,
+      "loss": 0.5409,
+      "step": 7942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016975291926187e-05,
+      "loss": 0.5002,
+      "step": 7943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701617676536968e-05,
+      "loss": 0.5607,
+      "step": 7944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015378150689604e-05,
+      "loss": 0.5512,
+      "step": 7945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7014579447895983e-05,
+      "loss": 0.4928,
+      "step": 7946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013780656998854e-05,
+      "loss": 0.488,
+      "step": 7947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012981778008245e-05,
+      "loss": 0.5548,
+      "step": 7948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012182810934192e-05,
+      "loss": 0.5855,
+      "step": 7949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701138375578673e-05,
+      "loss": 0.5557,
+      "step": 7950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7010584612575897e-05,
+      "loss": 0.6073,
+      "step": 7951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700978538131173e-05,
+      "loss": 0.583,
+      "step": 7952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008986062004266e-05,
+      "loss": 0.5653,
+      "step": 7953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700818665466354e-05,
+      "loss": 0.6136,
+      "step": 7954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70073871592996e-05,
+      "loss": 0.5493,
+      "step": 7955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006587575922484e-05,
+      "loss": 0.5015,
+      "step": 7956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005787904542236e-05,
+      "loss": 0.5689,
+      "step": 7957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004988145168902e-05,
+      "loss": 0.5503,
+      "step": 7958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004188297812518e-05,
+      "loss": 0.6433,
+      "step": 7959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7003388362483143e-05,
+      "loss": 0.5662,
+      "step": 7960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7002588339190816e-05,
+      "loss": 0.5689,
+      "step": 7961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001788227945586e-05,
+      "loss": 0.5319,
+      "step": 7962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000988028757506e-05,
+      "loss": 0.5642,
+      "step": 7963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000187741636622e-05,
+      "loss": 0.5481,
+      "step": 7964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699938736659299e-05,
+      "loss": 0.5581,
+      "step": 7965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699858690363666e-05,
+      "loss": 0.5307,
+      "step": 7966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6997786352777686e-05,
+      "loss": 0.5791,
+      "step": 7967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699698571402612e-05,
+      "loss": 0.5566,
+      "step": 7968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699618498739203e-05,
+      "loss": 0.5499,
+      "step": 7969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699538417288546e-05,
+      "loss": 0.5807,
+      "step": 7970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699458327051647e-05,
+      "loss": 0.5193,
+      "step": 7971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699378228029513e-05,
+      "loss": 0.6108,
+      "step": 7972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992981202231487e-05,
+      "loss": 0.4891,
+      "step": 7973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992180036335608e-05,
+      "loss": 0.5262,
+      "step": 7974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991378782617556e-05,
+      "loss": 0.4903,
+      "step": 7975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69905774410874e-05,
+      "loss": 0.6074,
+      "step": 7976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989776011755193e-05,
+      "loss": 0.5708,
+      "step": 7977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988974494631017e-05,
+      "loss": 0.5365,
+      "step": 7978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988172889724923e-05,
+      "loss": 0.49,
+      "step": 7979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6987371197046986e-05,
+      "loss": 0.5497,
+      "step": 7980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698656941660728e-05,
+      "loss": 0.5995,
+      "step": 7981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985767548415865e-05,
+      "loss": 0.5442,
+      "step": 7982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698496559248282e-05,
+      "loss": 0.537,
+      "step": 7983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984163548818218e-05,
+      "loss": 0.5331,
+      "step": 7984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983361417432128e-05,
+      "loss": 0.5304,
+      "step": 7985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698255919833463e-05,
+      "loss": 0.5792,
+      "step": 7986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981756891535798e-05,
+      "loss": 0.4985,
+      "step": 7987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6980954497045706e-05,
+      "loss": 0.5808,
+      "step": 7988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6980152014874436e-05,
+      "loss": 0.6175,
+      "step": 7989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979349445032065e-05,
+      "loss": 0.5572,
+      "step": 7990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6978546787528672e-05,
+      "loss": 0.5623,
+      "step": 7991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977744042374347e-05,
+      "loss": 0.5334,
+      "step": 7992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976941209579162e-05,
+      "loss": 0.5392,
+      "step": 7993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976138289153205e-05,
+      "loss": 0.5959,
+      "step": 7994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697533528110656e-05,
+      "loss": 0.5653,
+      "step": 7995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697453218544931e-05,
+      "loss": 0.5278,
+      "step": 7996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697372900219155e-05,
+      "loss": 0.558,
+      "step": 7997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697292573134336e-05,
+      "loss": 0.5475,
+      "step": 7998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972122372914834e-05,
+      "loss": 0.5411,
+      "step": 7999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971318926916063e-05,
+      "loss": 0.5645,
+      "step": 8000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697051539335713e-05,
+      "loss": 0.4601,
+      "step": 8001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6969711772248135e-05,
+      "loss": 0.5716,
+      "step": 8002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968908063599175e-05,
+      "loss": 0.5035,
+      "step": 8003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968104267420335e-05,
+      "loss": 0.5873,
+      "step": 8004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967300383721713e-05,
+      "loss": 0.5518,
+      "step": 8005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966496412513414e-05,
+      "loss": 0.5286,
+      "step": 8006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965692353805527e-05,
+      "loss": 0.5468,
+      "step": 8007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964888207608153e-05,
+      "loss": 0.5564,
+      "step": 8008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964083973931394e-05,
+      "loss": 0.539,
+      "step": 8009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6963279652785348e-05,
+      "loss": 0.5616,
+      "step": 8010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962475244180124e-05,
+      "loss": 0.553,
+      "step": 8011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961670748125817e-05,
+      "loss": 0.6447,
+      "step": 8012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6960866164632538e-05,
+      "loss": 0.5833,
+      "step": 8013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696006149371039e-05,
+      "loss": 0.5418,
+      "step": 8014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959256735369476e-05,
+      "loss": 0.5613,
+      "step": 8015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695845188961991e-05,
+      "loss": 0.4996,
+      "step": 8016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69576469564718e-05,
+      "loss": 0.528,
+      "step": 8017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956841935935254e-05,
+      "loss": 0.5582,
+      "step": 8018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956036828020385e-05,
+      "loss": 0.6047,
+      "step": 8019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69552316327373e-05,
+      "loss": 0.5637,
+      "step": 8020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954426350096118e-05,
+      "loss": 0.5026,
+      "step": 8021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695362098010695e-05,
+      "loss": 0.6109,
+      "step": 8022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6952815522779915e-05,
+      "loss": 0.4921,
+      "step": 8023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695200997812513e-05,
+      "loss": 0.6348,
+      "step": 8024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951204346152708e-05,
+      "loss": 0.4989,
+      "step": 8025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6950398626872768e-05,
+      "loss": 0.5531,
+      "step": 8026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949592820295437e-05,
+      "loss": 0.6745,
+      "step": 8027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694878692643083e-05,
+      "loss": 0.5832,
+      "step": 8028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947980945289064e-05,
+      "loss": 0.5429,
+      "step": 8029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947174876880272e-05,
+      "loss": 0.5725,
+      "step": 8030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946368721214575e-05,
+      "loss": 0.535,
+      "step": 8031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69455624783021e-05,
+      "loss": 0.5042,
+      "step": 8032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944756148152965e-05,
+      "loss": 0.5478,
+      "step": 8033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943949730777305e-05,
+      "loss": 0.512,
+      "step": 8034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943143226185252e-05,
+      "loss": 0.5205,
+      "step": 8035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694233663438693e-05,
+      "loss": 0.5476,
+      "step": 8036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694152995539247e-05,
+      "loss": 0.5972,
+      "step": 8037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6940723189212003e-05,
+      "loss": 0.6104,
+      "step": 8038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693991633585567e-05,
+      "loss": 0.5426,
+      "step": 8039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939109395333593e-05,
+      "loss": 0.5312,
+      "step": 8040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938302367655917e-05,
+      "loss": 0.545,
+      "step": 8041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6937495252832773e-05,
+      "loss": 0.5884,
+      "step": 8042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69366880508743e-05,
+      "loss": 0.6206,
+      "step": 8043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6935880761790638e-05,
+      "loss": 0.5051,
+      "step": 8044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693507338559192e-05,
+      "loss": 0.5052,
+      "step": 8045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69342659222883e-05,
+      "loss": 0.5501,
+      "step": 8046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693345837188991e-05,
+      "loss": 0.5531,
+      "step": 8047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693265073440689e-05,
+      "loss": 0.5991,
+      "step": 8048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931843009849388e-05,
+      "loss": 0.5269,
+      "step": 8049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931035198227553e-05,
+      "loss": 0.5707,
+      "step": 8050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930227299551528e-05,
+      "loss": 0.5261,
+      "step": 8051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692941931383146e-05,
+      "loss": 0.5845,
+      "step": 8052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69286112410775e-05,
+      "loss": 0.5466,
+      "step": 8053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927803081299787e-05,
+      "loss": 0.5339,
+      "step": 8054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926994834508487e-05,
+      "loss": 0.5379,
+      "step": 8055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926186500713737e-05,
+      "loss": 0.5705,
+      "step": 8056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69253780799257e-05,
+      "loss": 0.5495,
+      "step": 8057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692456957215453e-05,
+      "loss": 0.5118,
+      "step": 8058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6923760977410372e-05,
+      "loss": 0.5697,
+      "step": 8059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692295229570339e-05,
+      "loss": 0.5416,
+      "step": 8060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692214352704374e-05,
+      "loss": 0.5352,
+      "step": 8061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921334671441584e-05,
+      "loss": 0.5926,
+      "step": 8062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692052572890707e-05,
+      "loss": 0.586,
+      "step": 8063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691971669945037e-05,
+      "loss": 0.5592,
+      "step": 8064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691890758308164e-05,
+      "loss": 0.5918,
+      "step": 8065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691809837981104e-05,
+      "loss": 0.5397,
+      "step": 8066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6917289089648742e-05,
+      "loss": 0.5286,
+      "step": 8067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69164797126049e-05,
+      "loss": 0.5525,
+      "step": 8068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691567024868969e-05,
+      "loss": 0.5554,
+      "step": 8069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691486069791327e-05,
+      "loss": 0.5767,
+      "step": 8070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914051060285815e-05,
+      "loss": 0.6293,
+      "step": 8071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6913241335817492e-05,
+      "loss": 0.5846,
+      "step": 8072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691243152451847e-05,
+      "loss": 0.5762,
+      "step": 8073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911621626398923e-05,
+      "loss": 0.5458,
+      "step": 8074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691081164146902e-05,
+      "loss": 0.5751,
+      "step": 8075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910001569738933e-05,
+      "loss": 0.5137,
+      "step": 8076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909191411218842e-05,
+      "loss": 0.5466,
+      "step": 8077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690838116591892e-05,
+      "loss": 0.4662,
+      "step": 8078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907570833849342e-05,
+      "loss": 0.5936,
+      "step": 8079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690676041502029e-05,
+      "loss": 0.5605,
+      "step": 8080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690594990944194e-05,
+      "loss": 0.5474,
+      "step": 8081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690513931712447e-05,
+      "loss": 0.568,
+      "step": 8082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690432863807806e-05,
+      "loss": 0.5649,
+      "step": 8083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69035178723129e-05,
+      "loss": 0.5386,
+      "step": 8084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6902707019839174e-05,
+      "loss": 0.5638,
+      "step": 8085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901896080667055e-05,
+      "loss": 0.58,
+      "step": 8086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901085054806734e-05,
+      "loss": 0.5435,
+      "step": 8087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900273942268403e-05,
+      "loss": 0.5492,
+      "step": 8088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689946274306224e-05,
+      "loss": 0.5682,
+      "step": 8089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689865145719844e-05,
+      "loss": 0.6333,
+      "step": 8090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897840084687193e-05,
+      "loss": 0.5856,
+      "step": 8091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897028625538688e-05,
+      "loss": 0.5759,
+      "step": 8092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896217079763116e-05,
+      "loss": 0.576,
+      "step": 8093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689540544737067e-05,
+      "loss": 0.5416,
+      "step": 8094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689459372837155e-05,
+      "loss": 0.483,
+      "step": 8095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689378192277594e-05,
+      "loss": 0.5782,
+      "step": 8096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892970030594048e-05,
+      "loss": 0.5652,
+      "step": 8097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892158051836064e-05,
+      "loss": 0.5393,
+      "step": 8098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689134598651219e-05,
+      "loss": 0.4894,
+      "step": 8099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890533834632626e-05,
+      "loss": 0.5592,
+      "step": 8100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889721596207568e-05,
+      "loss": 0.547,
+      "step": 8101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888909271247222e-05,
+      "loss": 0.4947,
+      "step": 8102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688809685976179e-05,
+      "loss": 0.5393,
+      "step": 8103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887284361761476e-05,
+      "loss": 0.6837,
+      "step": 8104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886471777256483e-05,
+      "loss": 0.5548,
+      "step": 8105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688565910625702e-05,
+      "loss": 0.5794,
+      "step": 8106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884846348773294e-05,
+      "loss": 0.5703,
+      "step": 8107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884033504815512e-05,
+      "loss": 0.521,
+      "step": 8108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883220574393885e-05,
+      "loss": 0.6049,
+      "step": 8109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882407557518616e-05,
+      "loss": 0.5212,
+      "step": 8110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688159445419993e-05,
+      "loss": 0.5424,
+      "step": 8111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880781264448027e-05,
+      "loss": 0.5712,
+      "step": 8112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6879967988273127e-05,
+      "loss": 0.5963,
+      "step": 8113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6879154625685444e-05,
+      "loss": 0.5525,
+      "step": 8114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6878341176695196e-05,
+      "loss": 0.5492,
+      "step": 8115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6877527641312595e-05,
+      "loss": 0.5162,
+      "step": 8116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876714019547862e-05,
+      "loss": 0.5103,
+      "step": 8117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875900311411215e-05,
+      "loss": 0.6632,
+      "step": 8118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875086516912876e-05,
+      "loss": 0.5703,
+      "step": 8119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874272636063066e-05,
+      "loss": 0.5361,
+      "step": 8120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873458668872007e-05,
+      "loss": 0.5211,
+      "step": 8121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687264461534992e-05,
+      "loss": 0.4556,
+      "step": 8122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871830475507033e-05,
+      "loss": 0.6109,
+      "step": 8123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871016249353573e-05,
+      "loss": 0.5455,
+      "step": 8124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687020193689976e-05,
+      "loss": 0.5395,
+      "step": 8125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686938753815583e-05,
+      "loss": 0.5472,
+      "step": 8126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868573053132e-05,
+      "loss": 0.5082,
+      "step": 8127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867758481838517e-05,
+      "loss": 0.5862,
+      "step": 8128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6866943824285597e-05,
+      "loss": 0.5474,
+      "step": 8129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686612908048348e-05,
+      "loss": 0.5643,
+      "step": 8130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68653142504424e-05,
+      "loss": 0.5955,
+      "step": 8131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864499334172585e-05,
+      "loss": 0.567,
+      "step": 8132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686368433168427e-05,
+      "loss": 0.6046,
+      "step": 8133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68628692429877e-05,
+      "loss": 0.5556,
+      "step": 8134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686205406809311e-05,
+      "loss": 0.5733,
+      "step": 8135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861238807010733e-05,
+      "loss": 0.5682,
+      "step": 8136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860423459750817e-05,
+      "loss": 0.5833,
+      "step": 8137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6859608026323594e-05,
+      "loss": 0.488,
+      "step": 8138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.685879250673931e-05,
+      "loss": 0.5603,
+      "step": 8139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685797690100821e-05,
+      "loss": 0.5508,
+      "step": 8140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685716120914053e-05,
+      "loss": 0.5372,
+      "step": 8141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685634543114653e-05,
+      "loss": 0.5748,
+      "step": 8142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855529567036443e-05,
+      "loss": 0.5844,
+      "step": 8143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685471361682052e-05,
+      "loss": 0.5564,
+      "step": 8144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6853897580509012e-05,
+      "loss": 0.5349,
+      "step": 8145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6853081458112167e-05,
+      "loss": 0.6596,
+      "step": 8146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852265249640232e-05,
+      "loss": 0.5804,
+      "step": 8147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851448955103462e-05,
+      "loss": 0.57,
+      "step": 8148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850632574512113e-05,
+      "loss": 0.5441,
+      "step": 8149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849816107876432e-05,
+      "loss": 0.5088,
+      "step": 8150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848999555206676e-05,
+      "loss": 0.5549,
+      "step": 8151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848182916513108e-05,
+      "loss": 0.4981,
+      "step": 8152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684736619180597e-05,
+      "loss": 0.5348,
+      "step": 8153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846549381095535e-05,
+      "loss": 0.5227,
+      "step": 8154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845732484392055e-05,
+      "loss": 0.5537,
+      "step": 8155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844915501705793e-05,
+      "loss": 0.5608,
+      "step": 8156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684409843304701e-05,
+      "loss": 0.5108,
+      "step": 8157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843281278425964e-05,
+      "loss": 0.5561,
+      "step": 8158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842464037852925e-05,
+      "loss": 0.5037,
+      "step": 8159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684164671133815e-05,
+      "loss": 0.5464,
+      "step": 8160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840829298891912e-05,
+      "loss": 0.6446,
+      "step": 8161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6840011800524477e-05,
+      "loss": 0.5298,
+      "step": 8162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683919421624611e-05,
+      "loss": 0.4602,
+      "step": 8163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838376546067076e-05,
+      "loss": 0.5603,
+      "step": 8164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837558789997653e-05,
+      "loss": 0.5436,
+      "step": 8165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836740948048108e-05,
+      "loss": 0.5608,
+      "step": 8166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835923020228714e-05,
+      "loss": 0.596,
+      "step": 8167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835105006549744e-05,
+      "loss": 0.521,
+      "step": 8168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6834286907021475e-05,
+      "loss": 0.493,
+      "step": 8169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833468721654175e-05,
+      "loss": 0.5858,
+      "step": 8170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832650450458124e-05,
+      "loss": 0.502,
+      "step": 8171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683183209344361e-05,
+      "loss": 0.5355,
+      "step": 8172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831013650620892e-05,
+      "loss": 0.5162,
+      "step": 8173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830195122000266e-05,
+      "loss": 0.5819,
+      "step": 8174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6829376507592005e-05,
+      "loss": 0.6134,
+      "step": 8175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682855780740639e-05,
+      "loss": 0.5423,
+      "step": 8176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682773902145371e-05,
+      "loss": 0.6089,
+      "step": 8177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6826920149744242e-05,
+      "loss": 0.5492,
+      "step": 8178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682610119228828e-05,
+      "loss": 0.4841,
+      "step": 8179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825282149096102e-05,
+      "loss": 0.5612,
+      "step": 8180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6824463020177995e-05,
+      "loss": 0.5347,
+      "step": 8181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823643805544257e-05,
+      "loss": 0.5859,
+      "step": 8182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822824505205164e-05,
+      "loss": 0.7187,
+      "step": 8183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822005119171016e-05,
+      "loss": 0.5366,
+      "step": 8184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821185647452103e-05,
+      "loss": 0.6304,
+      "step": 8185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6820366090058714e-05,
+      "loss": 0.5506,
+      "step": 8186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819546447001146e-05,
+      "loss": 0.5386,
+      "step": 8187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818726718289692e-05,
+      "loss": 0.5425,
+      "step": 8188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681790690393465e-05,
+      "loss": 0.5535,
+      "step": 8189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6817087003946313e-05,
+      "loss": 0.5496,
+      "step": 8190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681626701833498e-05,
+      "loss": 0.5426,
+      "step": 8191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6815446947110955e-05,
+      "loss": 0.5738,
+      "step": 8192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814626790284533e-05,
+      "loss": 0.523,
+      "step": 8193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813806547866017e-05,
+      "loss": 0.6386,
+      "step": 8194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812986219865712e-05,
+      "loss": 0.6002,
+      "step": 8195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812165806293914e-05,
+      "loss": 0.5951,
+      "step": 8196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811345307160936e-05,
+      "loss": 0.6213,
+      "step": 8197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810524722477078e-05,
+      "loss": 0.59,
+      "step": 8198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6809704052252646e-05,
+      "loss": 0.562,
+      "step": 8199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808883296497952e-05,
+      "loss": 0.5454,
+      "step": 8200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68080624552233e-05,
+      "loss": 0.5903,
+      "step": 8201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807241528439003e-05,
+      "loss": 0.5502,
+      "step": 8202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806420516155376e-05,
+      "loss": 0.5928,
+      "step": 8203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680559941838272e-05,
+      "loss": 0.5365,
+      "step": 8204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6804778235131357e-05,
+      "loss": 0.5707,
+      "step": 8205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68039569664116e-05,
+      "loss": 0.5123,
+      "step": 8206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680313561223376e-05,
+      "loss": 0.5965,
+      "step": 8207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802314172608155e-05,
+      "loss": 0.5079,
+      "step": 8208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801492647545107e-05,
+      "loss": 0.5903,
+      "step": 8209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800671037054924e-05,
+      "loss": 0.5949,
+      "step": 8210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679984934114794e-05,
+      "loss": 0.6186,
+      "step": 8211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679902755983446e-05,
+      "loss": 0.5805,
+      "step": 8212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798205693124816e-05,
+      "loss": 0.5305,
+      "step": 8213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6797383741029328e-05,
+      "loss": 0.5559,
+      "step": 8214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796561703558317e-05,
+      "loss": 0.6298,
+      "step": 8215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679573958072211e-05,
+      "loss": 0.6251,
+      "step": 8216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6794917372531037e-05,
+      "loss": 0.6197,
+      "step": 8217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679409507899542e-05,
+      "loss": 0.5548,
+      "step": 8218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679327270012559e-05,
+      "loss": 0.5787,
+      "step": 8219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679245023593187e-05,
+      "loss": 0.5168,
+      "step": 8220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791627686424598e-05,
+      "loss": 0.6327,
+      "step": 8221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67908050516141e-05,
+      "loss": 0.5157,
+      "step": 8222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789982331510716e-05,
+      "loss": 0.5616,
+      "step": 8223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789159526124764e-05,
+      "loss": 0.4947,
+      "step": 8224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788336635466596e-05,
+      "loss": 0.5683,
+      "step": 8225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787513659546535e-05,
+      "loss": 0.5316,
+      "step": 8226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786690598374925e-05,
+      "loss": 0.5743,
+      "step": 8227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785867451962098e-05,
+      "loss": 0.5236,
+      "step": 8228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67850442203184e-05,
+      "loss": 0.5829,
+      "step": 8229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6784220903454163e-05,
+      "loss": 0.5344,
+      "step": 8230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6783397501379734e-05,
+      "loss": 0.6142,
+      "step": 8231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678257401410545e-05,
+      "loss": 0.5238,
+      "step": 8232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781750441641656e-05,
+      "loss": 0.5806,
+      "step": 8233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6780926783998702e-05,
+      "loss": 0.5292,
+      "step": 8234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678010304118692e-05,
+      "loss": 0.5056,
+      "step": 8235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.677927921321667e-05,
+      "loss": 0.5233,
+      "step": 8236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778455300098297e-05,
+      "loss": 0.5196,
+      "step": 8237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777631301842137e-05,
+      "loss": 0.5902,
+      "step": 8238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776807218458553e-05,
+      "loss": 0.5238,
+      "step": 8239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6775983049957888e-05,
+      "loss": 0.5705,
+      "step": 8240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67751587963505e-05,
+      "loss": 0.5474,
+      "step": 8241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774334457646734e-05,
+      "loss": 0.6122,
+      "step": 8242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773510033856952e-05,
+      "loss": 0.5512,
+      "step": 8243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67726855249915e-05,
+      "loss": 0.5122,
+      "step": 8244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771860931060745e-05,
+      "loss": 0.5585,
+      "step": 8245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677103625207503e-05,
+      "loss": 0.5211,
+      "step": 8246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770211488044722e-05,
+      "loss": 0.5652,
+      "step": 8247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769386638980182e-05,
+      "loss": 0.56,
+      "step": 8248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768561704891766e-05,
+      "loss": 0.5794,
+      "step": 8249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767736685789835e-05,
+      "loss": 0.5166,
+      "step": 8250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766911581684752e-05,
+      "loss": 0.5919,
+      "step": 8251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766086392586883e-05,
+      "loss": 0.5791,
+      "step": 8252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765261118506588e-05,
+      "loss": 0.5149,
+      "step": 8253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676443575945423e-05,
+      "loss": 0.5279,
+      "step": 8254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763610315440186e-05,
+      "loss": 0.5624,
+      "step": 8255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762784786474816e-05,
+      "loss": 0.6261,
+      "step": 8256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6761959172568493e-05,
+      "loss": 0.5477,
+      "step": 8257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676113347373158e-05,
+      "loss": 0.5583,
+      "step": 8258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676030768997445e-05,
+      "loss": 0.5827,
+      "step": 8259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675948182130748e-05,
+      "loss": 0.6702,
+      "step": 8260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675865586774104e-05,
+      "loss": 0.5853,
+      "step": 8261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757829829285503e-05,
+      "loss": 0.5426,
+      "step": 8262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675700370595125e-05,
+      "loss": 0.5595,
+      "step": 8263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756177497748643e-05,
+      "loss": 0.6504,
+      "step": 8264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675535120468807e-05,
+      "loss": 0.5398,
+      "step": 8265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754524826779914e-05,
+      "loss": 0.5812,
+      "step": 8266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753698364034543e-05,
+      "loss": 0.5588,
+      "step": 8267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752871816462343e-05,
+      "loss": 0.5767,
+      "step": 8268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752045184073695e-05,
+      "loss": 0.5582,
+      "step": 8269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675121846687898e-05,
+      "loss": 0.5236,
+      "step": 8270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750391664888585e-05,
+      "loss": 0.5681,
+      "step": 8271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749564778112893e-05,
+      "loss": 0.5885,
+      "step": 8272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6748737806562288e-05,
+      "loss": 0.613,
+      "step": 8273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6747910750247158e-05,
+      "loss": 0.558,
+      "step": 8274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674708360917789e-05,
+      "loss": 0.6154,
+      "step": 8275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746256383364875e-05,
+      "loss": 0.5695,
+      "step": 8276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6745429072818508e-05,
+      "loss": 0.5681,
+      "step": 8277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744601677549168e-05,
+      "loss": 0.5801,
+      "step": 8278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743774197567253e-05,
+      "loss": 0.5864,
+      "step": 8279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742946632883158e-05,
+      "loss": 0.6132,
+      "step": 8280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674211898350727e-05,
+      "loss": 0.5375,
+      "step": 8281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6741291249449997e-05,
+      "loss": 0.6086,
+      "step": 8282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740463430721728e-05,
+      "loss": 0.5896,
+      "step": 8283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739635527332857e-05,
+      "loss": 0.5003,
+      "step": 8284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738807539293788e-05,
+      "loss": 0.5018,
+      "step": 8285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673797946661492e-05,
+      "loss": 0.5883,
+      "step": 8286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6737151309306653e-05,
+      "loss": 0.5293,
+      "step": 8287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736323067379388e-05,
+      "loss": 0.5309,
+      "step": 8288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735494740843523e-05,
+      "loss": 0.605,
+      "step": 8289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734666329709473e-05,
+      "loss": 0.5461,
+      "step": 8290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733837833987634e-05,
+      "loss": 0.6219,
+      "step": 8291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733009253688416e-05,
+      "loss": 0.6047,
+      "step": 8292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673218058882222e-05,
+      "loss": 0.5746,
+      "step": 8293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731351839399466e-05,
+      "loss": 0.5527,
+      "step": 8294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730523005430553e-05,
+      "loss": 0.5464,
+      "step": 8295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729694086925893e-05,
+      "loss": 0.4951,
+      "step": 8296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67288650838959e-05,
+      "loss": 0.5593,
+      "step": 8297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672803599635098e-05,
+      "loss": 0.6025,
+      "step": 8298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727206824301555e-05,
+      "loss": 0.5288,
+      "step": 8299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726377567758036e-05,
+      "loss": 0.5169,
+      "step": 8300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725548226730837e-05,
+      "loss": 0.562,
+      "step": 8301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724718801230374e-05,
+      "loss": 0.5838,
+      "step": 8302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672388929126707e-05,
+      "loss": 0.5193,
+      "step": 8303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672305969685134e-05,
+      "loss": 0.5737,
+      "step": 8304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67222300179936e-05,
+      "loss": 0.5562,
+      "step": 8305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6721400254704273e-05,
+      "loss": 0.5995,
+      "step": 8306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6720570406993787e-05,
+      "loss": 0.554,
+      "step": 8307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671974047487256e-05,
+      "loss": 0.6028,
+      "step": 8308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718910458351016e-05,
+      "loss": 0.6136,
+      "step": 8309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671808035743958e-05,
+      "loss": 0.5341,
+      "step": 8310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671725017214868e-05,
+      "loss": 0.5516,
+      "step": 8311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6716419902488733e-05,
+      "loss": 0.6191,
+      "step": 8312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715589548470187e-05,
+      "loss": 0.5845,
+      "step": 8313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714759110103457e-05,
+      "loss": 0.5151,
+      "step": 8314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671392858739897e-05,
+      "loss": 0.6003,
+      "step": 8315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671309798036717e-05,
+      "loss": 0.5645,
+      "step": 8316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6712267289018486e-05,
+      "loss": 0.5517,
+      "step": 8317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671143651336334e-05,
+      "loss": 0.4788,
+      "step": 8318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6710605653412186e-05,
+      "loss": 0.5542,
+      "step": 8319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709774709175444e-05,
+      "loss": 0.6545,
+      "step": 8320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708943680663553e-05,
+      "loss": 0.5249,
+      "step": 8321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708112567886958e-05,
+      "loss": 0.5342,
+      "step": 8322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670728137085609e-05,
+      "loss": 0.4763,
+      "step": 8323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706450089581395e-05,
+      "loss": 0.5062,
+      "step": 8324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6705618724073312e-05,
+      "loss": 0.56,
+      "step": 8325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6704787274342277e-05,
+      "loss": 0.5132,
+      "step": 8326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703955740398745e-05,
+      "loss": 0.563,
+      "step": 8327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703124122253148e-05,
+      "loss": 0.5321,
+      "step": 8328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670229241991594e-05,
+      "loss": 0.5012,
+      "step": 8329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6701460633397563e-05,
+      "loss": 0.6396,
+      "step": 8330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700628762708467e-05,
+      "loss": 0.6029,
+      "step": 8331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6699796807859094e-05,
+      "loss": 0.5326,
+      "step": 8332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66989647688599e-05,
+      "loss": 0.5515,
+      "step": 8333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698132645721337e-05,
+      "loss": 0.5076,
+      "step": 8334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6697300438453846e-05,
+      "loss": 0.5709,
+      "step": 8335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669646814706789e-05,
+      "loss": 0.5179,
+      "step": 8336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695635771573922e-05,
+      "loss": 0.5223,
+      "step": 8337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694803311982387e-05,
+      "loss": 0.5116,
+      "step": 8338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693970768303753e-05,
+      "loss": 0.5627,
+      "step": 8339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669313814054847e-05,
+      "loss": 0.5926,
+      "step": 8340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6692305428726997e-05,
+      "loss": 0.4583,
+      "step": 8341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691472632849792e-05,
+      "loss": 0.5362,
+      "step": 8342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690639752927315e-05,
+      "loss": 0.5163,
+      "step": 8343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6689806788970032e-05,
+      "loss": 0.4966,
+      "step": 8344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66889737409884e-05,
+      "loss": 0.6043,
+      "step": 8345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688140608992877e-05,
+      "loss": 0.4872,
+      "step": 8346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687307392993937e-05,
+      "loss": 0.5162,
+      "step": 8347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6686474093002045e-05,
+      "loss": 0.5522,
+      "step": 8348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685640709027663e-05,
+      "loss": 0.5295,
+      "step": 8349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668480724108126e-05,
+      "loss": 0.5747,
+      "step": 8350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6683973689173298e-05,
+      "loss": 0.5193,
+      "step": 8351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6683140053314255e-05,
+      "loss": 0.565,
+      "step": 8352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6682306333514603e-05,
+      "loss": 0.5155,
+      "step": 8353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6681472529784803e-05,
+      "loss": 0.5304,
+      "step": 8354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680638642135337e-05,
+      "loss": 0.4703,
+      "step": 8355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679804670576672e-05,
+      "loss": 0.5123,
+      "step": 8356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678970615119292e-05,
+      "loss": 0.5552,
+      "step": 8357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678136475773664e-05,
+      "loss": 0.5457,
+      "step": 8358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667730225255027e-05,
+      "loss": 0.5567,
+      "step": 8359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667646794545958e-05,
+      "loss": 0.533,
+      "step": 8360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675633554512084e-05,
+      "loss": 0.5943,
+      "step": 8361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674799079718254e-05,
+      "loss": 0.5304,
+      "step": 8362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673964521088572e-05,
+      "loss": 0.5606,
+      "step": 8363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673129878633528e-05,
+      "loss": 0.5663,
+      "step": 8364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6672295152363595e-05,
+      "loss": 0.5512,
+      "step": 8365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667146034228926e-05,
+      "loss": 0.5139,
+      "step": 8366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670625448421012e-05,
+      "loss": 0.5486,
+      "step": 8367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669790470769335e-05,
+      "loss": 0.5514,
+      "step": 8368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668955409344714e-05,
+      "loss": 0.5689,
+      "step": 8369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666812026415764e-05,
+      "loss": 0.6099,
+      "step": 8370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6667285035218604e-05,
+      "loss": 0.5467,
+      "step": 8371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6666449722538095e-05,
+      "loss": 0.5015,
+      "step": 8372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66656143261266e-05,
+      "loss": 0.5546,
+      "step": 8373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664778845994617e-05,
+      "loss": 0.5315,
+      "step": 8374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666394328215264e-05,
+      "loss": 0.5278,
+      "step": 8375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666310763461116e-05,
+      "loss": 0.4772,
+      "step": 8376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662271903380678e-05,
+      "loss": 0.5038,
+      "step": 8377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661436088471688e-05,
+      "loss": 0.5749,
+      "step": 8378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6660600189894688e-05,
+      "loss": 0.6184,
+      "step": 8379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665976420766017e-05,
+      "loss": 0.5171,
+      "step": 8380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665892814177865e-05,
+      "loss": 0.6211,
+      "step": 8381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658091992260613e-05,
+      "loss": 0.5136,
+      "step": 8382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665725575911657e-05,
+      "loss": 0.5858,
+      "step": 8383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656419442357018e-05,
+      "loss": 0.5773,
+      "step": 8384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6655583041992472e-05,
+      "loss": 0.5754,
+      "step": 8385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654746558033424e-05,
+      "loss": 0.5675,
+      "step": 8386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665390999049039e-05,
+      "loss": 0.5634,
+      "step": 8387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665307333937387e-05,
+      "loss": 0.5712,
+      "step": 8388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665223660469438e-05,
+      "loss": 0.5128,
+      "step": 8389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651399786462426e-05,
+      "loss": 0.5609,
+      "step": 8390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650562884688517e-05,
+      "loss": 0.556,
+      "step": 8391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6649725899383167e-05,
+      "loss": 0.5812,
+      "step": 8392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648888830556886e-05,
+      "loss": 0.5462,
+      "step": 8393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664805167822019e-05,
+      "loss": 0.6338,
+      "step": 8394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647214442383587e-05,
+      "loss": 0.6074,
+      "step": 8395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646377123057605e-05,
+      "loss": 0.6306,
+      "step": 8396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664553972025275e-05,
+      "loss": 0.5235,
+      "step": 8397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6644702233979545e-05,
+      "loss": 0.5035,
+      "step": 8398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664386466424851e-05,
+      "loss": 0.4967,
+      "step": 8399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643027011070162e-05,
+      "loss": 0.5232,
+      "step": 8400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664218927445502e-05,
+      "loss": 0.5812,
+      "step": 8401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6641351454413612e-05,
+      "loss": 0.5663,
+      "step": 8402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640513550956456e-05,
+      "loss": 0.5989,
+      "step": 8403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663967556409408e-05,
+      "loss": 0.5187,
+      "step": 8404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638837493837002e-05,
+      "loss": 0.5565,
+      "step": 8405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663799934019576e-05,
+      "loss": 0.5381,
+      "step": 8406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663716110318087e-05,
+      "loss": 0.6033,
+      "step": 8407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6636322782802868e-05,
+      "loss": 0.6384,
+      "step": 8408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635484379072275e-05,
+      "loss": 0.5649,
+      "step": 8409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663464589199963e-05,
+      "loss": 0.6368,
+      "step": 8410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663380732159546e-05,
+      "loss": 0.5586,
+      "step": 8411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66329686678703e-05,
+      "loss": 0.5041,
+      "step": 8412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6632129930834683e-05,
+      "loss": 0.5612,
+      "step": 8413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663129111049914e-05,
+      "loss": 0.5007,
+      "step": 8414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663045220687421e-05,
+      "loss": 0.5918,
+      "step": 8415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662961321997043e-05,
+      "loss": 0.5451,
+      "step": 8416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6628774149798334e-05,
+      "loss": 0.541,
+      "step": 8417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627934996368464e-05,
+      "loss": 0.5124,
+      "step": 8418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627095759691364e-05,
+      "loss": 0.5277,
+      "step": 8419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626256439777565e-05,
+      "loss": 0.4572,
+      "step": 8420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625417036637616e-05,
+      "loss": 0.5228,
+      "step": 8421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624577550282056e-05,
+      "loss": 0.5215,
+      "step": 8422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6623737980721434e-05,
+      "loss": 0.5282,
+      "step": 8423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662289832796629e-05,
+      "loss": 0.6008,
+      "step": 8424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6622058592027173e-05,
+      "loss": 0.5407,
+      "step": 8425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621218772914625e-05,
+      "loss": 0.5168,
+      "step": 8426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620378870639204e-05,
+      "loss": 0.5584,
+      "step": 8427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6619538885211453e-05,
+      "loss": 0.5221,
+      "step": 8428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6618698816641922e-05,
+      "loss": 0.5414,
+      "step": 8429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617858664941163e-05,
+      "loss": 0.5431,
+      "step": 8430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617018430119725e-05,
+      "loss": 0.5339,
+      "step": 8431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661617811218817e-05,
+      "loss": 0.6346,
+      "step": 8432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615337711157046e-05,
+      "loss": 0.5734,
+      "step": 8433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661449722703691e-05,
+      "loss": 0.4952,
+      "step": 8434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613656659838316e-05,
+      "loss": 0.5499,
+      "step": 8435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612816009571826e-05,
+      "loss": 0.5365,
+      "step": 8436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6611975276247996e-05,
+      "loss": 0.4994,
+      "step": 8437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6611134459877386e-05,
+      "loss": 0.5683,
+      "step": 8438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661029356047056e-05,
+      "loss": 0.531,
+      "step": 8439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660945257803807e-05,
+      "loss": 0.6043,
+      "step": 8440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660861151259049e-05,
+      "loss": 0.592,
+      "step": 8441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660777036413838e-05,
+      "loss": 0.492,
+      "step": 8442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606929132692303e-05,
+      "loss": 0.5727,
+      "step": 8443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606087818262827e-05,
+      "loss": 0.51,
+      "step": 8444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6605246420860518e-05,
+      "loss": 0.5143,
+      "step": 8445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604404940495942e-05,
+      "loss": 0.5489,
+      "step": 8446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6603563377179672e-05,
+      "loss": 0.5785,
+      "step": 8447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602721730922274e-05,
+      "loss": 0.57,
+      "step": 8448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601880001734323e-05,
+      "loss": 0.6336,
+      "step": 8449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660103818962639e-05,
+      "loss": 0.536,
+      "step": 8450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6600196294609047e-05,
+      "loss": 0.5367,
+      "step": 8451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599354316692866e-05,
+      "loss": 0.6107,
+      "step": 8452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659851225588843e-05,
+      "loss": 0.4728,
+      "step": 8453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659767011220631e-05,
+      "loss": 0.5509,
+      "step": 8454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659682788565708e-05,
+      "loss": 0.5219,
+      "step": 8455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595985576251324e-05,
+      "loss": 0.4512,
+      "step": 8456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659514318399962e-05,
+      "loss": 0.5137,
+      "step": 8457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594300708912548e-05,
+      "loss": 0.593,
+      "step": 8458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.5301,
+      "step": 8459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659261551027463e-05,
+      "loss": 0.5848,
+      "step": 8460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659177278674495e-05,
+      "loss": 0.501,
+      "step": 8461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659092998042223e-05,
+      "loss": 0.6035,
+      "step": 8462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590087091317064e-05,
+      "loss": 0.5546,
+      "step": 8463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6589244119440032e-05,
+      "loss": 0.5323,
+      "step": 8464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658840106480173e-05,
+      "loss": 0.5395,
+      "step": 8465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587557927412738e-05,
+      "loss": 0.5444,
+      "step": 8466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658671470728365e-05,
+      "loss": 0.5226,
+      "step": 8467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585871404425056e-05,
+      "loss": 0.5631,
+      "step": 8468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658502801884755e-05,
+      "loss": 0.6155,
+      "step": 8469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658418455056172e-05,
+      "loss": 0.5621,
+      "step": 8470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583340999578167e-05,
+      "loss": 0.6172,
+      "step": 8471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658249736590748e-05,
+      "loss": 0.5767,
+      "step": 8472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581653649560262e-05,
+      "loss": 0.5503,
+      "step": 8473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65808098505471e-05,
+      "loss": 0.5337,
+      "step": 8474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65799659688786e-05,
+      "loss": 0.5269,
+      "step": 8475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657912200456536e-05,
+      "loss": 0.6079,
+      "step": 8476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6578277957617978e-05,
+      "loss": 0.5453,
+      "step": 8477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577433828047055e-05,
+      "loss": 0.5377,
+      "step": 8478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576589615863195e-05,
+      "loss": 0.6225,
+      "step": 8479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575745321077e-05,
+      "loss": 0.5163,
+      "step": 8480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657490094369908e-05,
+      "loss": 0.546,
+      "step": 8481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574056483740032e-05,
+      "loss": 0.532,
+      "step": 8482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573211941210467e-05,
+      "loss": 0.5671,
+      "step": 8483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572367316120994e-05,
+      "loss": 0.5116,
+      "step": 8484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571522608482217e-05,
+      "loss": 0.6144,
+      "step": 8485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570677818304748e-05,
+      "loss": 0.6009,
+      "step": 8486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65698329455992e-05,
+      "loss": 0.5439,
+      "step": 8487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6568987990376178e-05,
+      "loss": 0.635,
+      "step": 8488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6568142952646304e-05,
+      "loss": 0.5819,
+      "step": 8489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567297832420187e-05,
+      "loss": 0.6073,
+      "step": 8490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6566452629708437e-05,
+      "loss": 0.566,
+      "step": 8491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565607344521678e-05,
+      "loss": 0.5087,
+      "step": 8492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656476197687052e-05,
+      "loss": 0.5263,
+      "step": 8493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6563916526765585e-05,
+      "loss": 0.4595,
+      "step": 8494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656307099421749e-05,
+      "loss": 0.5644,
+      "step": 8495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6562225379236858e-05,
+      "loss": 0.4818,
+      "step": 8496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561379681834305e-05,
+      "loss": 0.545,
+      "step": 8497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560533902020458e-05,
+      "loss": 0.5164,
+      "step": 8498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6559688039805938e-05,
+      "loss": 0.5879,
+      "step": 8499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558842095201363e-05,
+      "loss": 0.5854,
+      "step": 8500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655799606821737e-05,
+      "loss": 0.5319,
+      "step": 8501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655714995886458e-05,
+      "loss": 0.6156,
+      "step": 8502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6556303767153618e-05,
+      "loss": 0.5769,
+      "step": 8503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655545749309511e-05,
+      "loss": 0.5907,
+      "step": 8504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655461113669969e-05,
+      "loss": 0.5764,
+      "step": 8505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655376469797799e-05,
+      "loss": 0.5363,
+      "step": 8506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552918176940637e-05,
+      "loss": 0.5706,
+      "step": 8507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655207157359826e-05,
+      "loss": 0.5124,
+      "step": 8508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551224887961502e-05,
+      "loss": 0.5356,
+      "step": 8509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6550378120040994e-05,
+      "loss": 0.5897,
+      "step": 8510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6549531269847366e-05,
+      "loss": 0.5207,
+      "step": 8511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654868433739126e-05,
+      "loss": 0.5483,
+      "step": 8512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6547837322683316e-05,
+      "loss": 0.5803,
+      "step": 8513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654699022573416e-05,
+      "loss": 0.5506,
+      "step": 8514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546143046554447e-05,
+      "loss": 0.582,
+      "step": 8515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654529578515481e-05,
+      "loss": 0.5353,
+      "step": 8516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544448441545887e-05,
+      "loss": 0.5074,
+      "step": 8517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543601015738328e-05,
+      "loss": 0.6045,
+      "step": 8518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6542753507742775e-05,
+      "loss": 0.5665,
+      "step": 8519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654190591756987e-05,
+      "loss": 0.6345,
+      "step": 8520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654105824523026e-05,
+      "loss": 0.5715,
+      "step": 8521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540210490734595e-05,
+      "loss": 0.5697,
+      "step": 8522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539362654093515e-05,
+      "loss": 0.535,
+      "step": 8523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538514735317676e-05,
+      "loss": 0.5879,
+      "step": 8524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6537666734417724e-05,
+      "loss": 0.5552,
+      "step": 8525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536818651404315e-05,
+      "loss": 0.6141,
+      "step": 8526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535970486288097e-05,
+      "loss": 0.5965,
+      "step": 8527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535122239079725e-05,
+      "loss": 0.5344,
+      "step": 8528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534273909789847e-05,
+      "loss": 0.5887,
+      "step": 8529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6533425498429126e-05,
+      "loss": 0.5119,
+      "step": 8530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532577005008213e-05,
+      "loss": 0.504,
+      "step": 8531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531728429537766e-05,
+      "loss": 0.5681,
+      "step": 8532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653087977202845e-05,
+      "loss": 0.5794,
+      "step": 8533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530031032490914e-05,
+      "loss": 0.5487,
+      "step": 8534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529182210935824e-05,
+      "loss": 0.5537,
+      "step": 8535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652833330737384e-05,
+      "loss": 0.5442,
+      "step": 8536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652748432181562e-05,
+      "loss": 0.5927,
+      "step": 8537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526635254271835e-05,
+      "loss": 0.5146,
+      "step": 8538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525786104753146e-05,
+      "loss": 0.5297,
+      "step": 8539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524936873270213e-05,
+      "loss": 0.5831,
+      "step": 8540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524087559833715e-05,
+      "loss": 0.5698,
+      "step": 8541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6523238164454305e-05,
+      "loss": 0.5365,
+      "step": 8542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6522388687142665e-05,
+      "loss": 0.5375,
+      "step": 8543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521539127909454e-05,
+      "loss": 0.5539,
+      "step": 8544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520689486765344e-05,
+      "loss": 0.5365,
+      "step": 8545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651983976372101e-05,
+      "loss": 0.5073,
+      "step": 8546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6518989958787126e-05,
+      "loss": 0.6804,
+      "step": 8547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651814007197436e-05,
+      "loss": 0.5766,
+      "step": 8548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517290103293392e-05,
+      "loss": 0.5629,
+      "step": 8549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6516440052754895e-05,
+      "loss": 0.6283,
+      "step": 8550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515589920369546e-05,
+      "loss": 0.56,
+      "step": 8551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514739706148024e-05,
+      "loss": 0.5137,
+      "step": 8552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513889410101004e-05,
+      "loss": 0.4786,
+      "step": 8553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651303903223917e-05,
+      "loss": 0.572,
+      "step": 8554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65121885725732e-05,
+      "loss": 0.563,
+      "step": 8555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511338031113776e-05,
+      "loss": 0.5582,
+      "step": 8556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510487407871584e-05,
+      "loss": 0.6005,
+      "step": 8557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509636702857305e-05,
+      "loss": 0.5528,
+      "step": 8558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508785916081622e-05,
+      "loss": 0.5494,
+      "step": 8559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507935047555222e-05,
+      "loss": 0.5924,
+      "step": 8560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507084097288798e-05,
+      "loss": 0.5842,
+      "step": 8561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650623306529303e-05,
+      "loss": 0.4452,
+      "step": 8562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505381951578613e-05,
+      "loss": 0.5409,
+      "step": 8563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504530756156236e-05,
+      "loss": 0.5462,
+      "step": 8564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650367947903658e-05,
+      "loss": 0.5443,
+      "step": 8565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502828120230354e-05,
+      "loss": 0.5142,
+      "step": 8566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650197667974824e-05,
+      "loss": 0.5923,
+      "step": 8567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6501125157600934e-05,
+      "loss": 0.527,
+      "step": 8568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6500273553799134e-05,
+      "loss": 0.6033,
+      "step": 8569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649942186835353e-05,
+      "loss": 0.6208,
+      "step": 8570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498570101274826e-05,
+      "loss": 0.5385,
+      "step": 8571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497718252573718e-05,
+      "loss": 0.5511,
+      "step": 8572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496866322260903e-05,
+      "loss": 0.57,
+      "step": 8573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496014310347086e-05,
+      "loss": 0.5675,
+      "step": 8574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495162216842962e-05,
+      "loss": 0.5521,
+      "step": 8575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649431004175924e-05,
+      "loss": 0.5251,
+      "step": 8576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493457785106615e-05,
+      "loss": 0.5357,
+      "step": 8577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64926054468958e-05,
+      "loss": 0.5841,
+      "step": 8578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6491753027137497e-05,
+      "loss": 0.5564,
+      "step": 8579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6490900525842416e-05,
+      "loss": 0.6221,
+      "step": 8580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6490047943021257e-05,
+      "loss": 0.5834,
+      "step": 8581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489195278684733e-05,
+      "loss": 0.5945,
+      "step": 8582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6488342532843552e-05,
+      "loss": 0.5187,
+      "step": 8583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6487489705508427e-05,
+      "loss": 0.556,
+      "step": 8584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486636796690067e-05,
+      "loss": 0.5793,
+      "step": 8585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485783806399185e-05,
+      "loss": 0.5085,
+      "step": 8586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484930734646496e-05,
+      "loss": 0.5836,
+      "step": 8587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648407758144271e-05,
+      "loss": 0.5579,
+      "step": 8588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483224346798552e-05,
+      "loss": 0.5596,
+      "step": 8589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648237103072473e-05,
+      "loss": 0.5756,
+      "step": 8590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481517633231966e-05,
+      "loss": 0.5412,
+      "step": 8591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648066415433098e-05,
+      "loss": 0.5868,
+      "step": 8592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479810594032485e-05,
+      "loss": 0.5175,
+      "step": 8593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478956952347205e-05,
+      "loss": 0.4788,
+      "step": 8594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478103229285868e-05,
+      "loss": 0.5622,
+      "step": 8595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6477249424859186e-05,
+      "loss": 0.5062,
+      "step": 8596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476395539077894e-05,
+      "loss": 0.5448,
+      "step": 8597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647554157195271e-05,
+      "loss": 0.593,
+      "step": 8598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647468752349436e-05,
+      "loss": 0.5789,
+      "step": 8599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647383339371357e-05,
+      "loss": 0.5331,
+      "step": 8600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472979182621075e-05,
+      "loss": 0.5643,
+      "step": 8601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472124890227596e-05,
+      "loss": 0.5825,
+      "step": 8602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6471270516543863e-05,
+      "loss": 0.5402,
+      "step": 8603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470416061580614e-05,
+      "loss": 0.5714,
+      "step": 8604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469561525348576e-05,
+      "loss": 0.5075,
+      "step": 8605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646870690785848e-05,
+      "loss": 0.4483,
+      "step": 8606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6467852209121068e-05,
+      "loss": 0.5518,
+      "step": 8607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466997429147066e-05,
+      "loss": 0.5768,
+      "step": 8608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466142567947215e-05,
+      "loss": 0.5828,
+      "step": 8609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6465287625532254e-05,
+      "loss": 0.5798,
+      "step": 8610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6464432601912914e-05,
+      "loss": 0.5919,
+      "step": 8611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463577497099943e-05,
+      "loss": 0.5705,
+      "step": 8612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462722311104072e-05,
+      "loss": 0.5352,
+      "step": 8613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461867043936052e-05,
+      "loss": 0.5505,
+      "step": 8614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461011695606615e-05,
+      "loss": 0.6392,
+      "step": 8615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646015626612651e-05,
+      "loss": 0.6135,
+      "step": 8616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459300755506484e-05,
+      "loss": 0.5676,
+      "step": 8617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6458445163757274e-05,
+      "loss": 0.612,
+      "step": 8618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457589490889633e-05,
+      "loss": 0.5992,
+      "step": 8619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645673373691431e-05,
+      "loss": 0.6202,
+      "step": 8620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6455877901842047e-05,
+      "loss": 0.596,
+      "step": 8621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6455021985683592e-05,
+      "loss": 0.5818,
+      "step": 8622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6454165988449706e-05,
+      "loss": 0.6069,
+      "step": 8623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645330991015113e-05,
+      "loss": 0.5637,
+      "step": 8624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645245375079862e-05,
+      "loss": 0.5706,
+      "step": 8625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645159751040293e-05,
+      "loss": 0.5058,
+      "step": 8626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6450741188974816e-05,
+      "loss": 0.5362,
+      "step": 8627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449884786525026e-05,
+      "loss": 0.5755,
+      "step": 8628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449028303064327e-05,
+      "loss": 0.5487,
+      "step": 8629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644817173860347e-05,
+      "loss": 0.5597,
+      "step": 8630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6447315093153213e-05,
+      "loss": 0.5644,
+      "step": 8631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644645836672432e-05,
+      "loss": 0.5449,
+      "step": 8632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445601559327548e-05,
+      "loss": 0.4743,
+      "step": 8633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444744670973662e-05,
+      "loss": 0.5113,
+      "step": 8634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443887701673418e-05,
+      "loss": 0.5909,
+      "step": 8635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443030651437583e-05,
+      "loss": 0.5103,
+      "step": 8636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442173520276926e-05,
+      "loss": 0.6138,
+      "step": 8637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441316308202206e-05,
+      "loss": 0.6103,
+      "step": 8638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644045901522419e-05,
+      "loss": 0.5645,
+      "step": 8639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6439601641353653e-05,
+      "loss": 0.5361,
+      "step": 8640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438744186601358e-05,
+      "loss": 0.5951,
+      "step": 8641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437886650978073e-05,
+      "loss": 0.5589,
+      "step": 8642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437029034494574e-05,
+      "loss": 0.5465,
+      "step": 8643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436171337161628e-05,
+      "loss": 0.5287,
+      "step": 8644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643531355899001e-05,
+      "loss": 0.568,
+      "step": 8645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643445569999049e-05,
+      "loss": 0.5518,
+      "step": 8646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433597760173852e-05,
+      "loss": 0.5587,
+      "step": 8647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432739739550862e-05,
+      "loss": 0.5441,
+      "step": 8648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64318816381323e-05,
+      "loss": 0.5791,
+      "step": 8649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6431023455928947e-05,
+      "loss": 0.5363,
+      "step": 8650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643016519295158e-05,
+      "loss": 0.5222,
+      "step": 8651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429306849210976e-05,
+      "loss": 0.5952,
+      "step": 8652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428448424717917e-05,
+      "loss": 0.5618,
+      "step": 8653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6427589919483187e-05,
+      "loss": 0.5511,
+      "step": 8654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6426731333517567e-05,
+      "loss": 0.5856,
+      "step": 8655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425872666831843e-05,
+      "loss": 0.5663,
+      "step": 8656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425013919436797e-05,
+      "loss": 0.5433,
+      "step": 8657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424155091343218e-05,
+      "loss": 0.5615,
+      "step": 8658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423296182561887e-05,
+      "loss": 0.5731,
+      "step": 8659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422437193103598e-05,
+      "loss": 0.5409,
+      "step": 8660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421578122979143e-05,
+      "loss": 0.5602,
+      "step": 8661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64207189721993e-05,
+      "loss": 0.512,
+      "step": 8662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419859740774873e-05,
+      "loss": 0.5611,
+      "step": 8663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419000428716646e-05,
+      "loss": 0.5444,
+      "step": 8664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641814103603541e-05,
+      "loss": 0.4846,
+      "step": 8665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641728156274197e-05,
+      "loss": 0.5941,
+      "step": 8666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641642200884711e-05,
+      "loss": 0.5681,
+      "step": 8667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415562374361632e-05,
+      "loss": 0.583,
+      "step": 8668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641470265929633e-05,
+      "loss": 0.6202,
+      "step": 8669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413842863662003e-05,
+      "loss": 0.5097,
+      "step": 8670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412982987469452e-05,
+      "loss": 0.5318,
+      "step": 8671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412123030729476e-05,
+      "loss": 0.5846,
+      "step": 8672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641126299345287e-05,
+      "loss": 0.6406,
+      "step": 8673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641040287565045e-05,
+      "loss": 0.5424,
+      "step": 8674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409542677333007e-05,
+      "loss": 0.5682,
+      "step": 8675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6408682398511348e-05,
+      "loss": 0.5792,
+      "step": 8676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407822039196283e-05,
+      "loss": 0.5696,
+      "step": 8677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640696159939861e-05,
+      "loss": 0.5346,
+      "step": 8678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640610107912914e-05,
+      "loss": 0.4787,
+      "step": 8679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6405240478398685e-05,
+      "loss": 0.5223,
+      "step": 8680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640437979721805e-05,
+      "loss": 0.5367,
+      "step": 8681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403519035598042e-05,
+      "loss": 0.5427,
+      "step": 8682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6402658193549485e-05,
+      "loss": 0.5852,
+      "step": 8683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6401797271083175e-05,
+      "loss": 0.5418,
+      "step": 8684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400936268209936e-05,
+      "loss": 0.5202,
+      "step": 8685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400075184940578e-05,
+      "loss": 0.5779,
+      "step": 8686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399214021285916e-05,
+      "loss": 0.4328,
+      "step": 8687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639835277725677e-05,
+      "loss": 0.5599,
+      "step": 8688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639749145286395e-05,
+      "loss": 0.5731,
+      "step": 8689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396630048118285e-05,
+      "loss": 0.5618,
+      "step": 8690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6395768563030585e-05,
+      "loss": 0.5292,
+      "step": 8691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394906997611676e-05,
+      "loss": 0.5081,
+      "step": 8692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394045351872373e-05,
+      "loss": 0.6009,
+      "step": 8693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393183625823508e-05,
+      "loss": 0.6029,
+      "step": 8694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392321819475892e-05,
+      "loss": 0.587,
+      "step": 8695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639145993284036e-05,
+      "loss": 0.4876,
+      "step": 8696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390597965927734e-05,
+      "loss": 0.5767,
+      "step": 8697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638973591874884e-05,
+      "loss": 0.5413,
+      "step": 8698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388873791314498e-05,
+      "loss": 0.5825,
+      "step": 8699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388011583635553e-05,
+      "loss": 0.6277,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638714929572282e-05,
+      "loss": 0.5772,
+      "step": 8701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6386286927587133e-05,
+      "loss": 0.5783,
+      "step": 8702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638542447923933e-05,
+      "loss": 0.5859,
+      "step": 8703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638456195069023e-05,
+      "loss": 0.5601,
+      "step": 8704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383699341950678e-05,
+      "loss": 0.5908,
+      "step": 8705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382836653031506e-05,
+      "loss": 0.5509,
+      "step": 8706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6381973883943546e-05,
+      "loss": 0.5464,
+      "step": 8707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638111103469764e-05,
+      "loss": 0.5405,
+      "step": 8708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380248105304617e-05,
+      "loss": 0.468,
+      "step": 8709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6379385095775323e-05,
+      "loss": 0.5816,
+      "step": 8710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378522006120595e-05,
+      "loss": 0.5188,
+      "step": 8711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377658836351275e-05,
+      "loss": 0.5321,
+      "step": 8712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63767955864782e-05,
+      "loss": 0.501,
+      "step": 8713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375932256512216e-05,
+      "loss": 0.617,
+      "step": 8714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637506884646417e-05,
+      "loss": 0.6313,
+      "step": 8715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374205356344898e-05,
+      "loss": 0.5858,
+      "step": 8716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6373341786165248e-05,
+      "loss": 0.5436,
+      "step": 8717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372478135936072e-05,
+      "loss": 0.5592,
+      "step": 8718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637161440566821e-05,
+      "loss": 0.5887,
+      "step": 8719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637075059537252e-05,
+      "loss": 0.5724,
+      "step": 8720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636988670505984e-05,
+      "loss": 0.4993,
+      "step": 8721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369022734741026e-05,
+      "loss": 0.5446,
+      "step": 8722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6368158684426933e-05,
+      "loss": 0.6036,
+      "step": 8723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636729455412841e-05,
+      "loss": 0.5571,
+      "step": 8724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366430343856313e-05,
+      "loss": 0.5563,
+      "step": 8725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636556605362149e-05,
+      "loss": 0.5508,
+      "step": 8726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364701683434804e-05,
+      "loss": 0.54,
+      "step": 8727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363837233307106e-05,
+      "loss": 0.5228,
+      "step": 8728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362972703249256e-05,
+      "loss": 0.5617,
+      "step": 8729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362108093272113e-05,
+      "loss": 0.5298,
+      "step": 8730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361243403386535e-05,
+      "loss": 0.5249,
+      "step": 8731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6360378633603386e-05,
+      "loss": 0.5702,
+      "step": 8732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359513783933523e-05,
+      "loss": 0.5558,
+      "step": 8733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6358648854387813e-05,
+      "loss": 0.5799,
+      "step": 8734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635778384497712e-05,
+      "loss": 0.5509,
+      "step": 8735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356918755712298e-05,
+      "loss": 0.5467,
+      "step": 8736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356053586604223e-05,
+      "loss": 0.574,
+      "step": 8737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635518833766376e-05,
+      "loss": 0.5656,
+      "step": 8738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354323008901774e-05,
+      "loss": 0.5458,
+      "step": 8739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635345760032914e-05,
+      "loss": 0.606,
+      "step": 8740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352592111956717e-05,
+      "loss": 0.578,
+      "step": 8741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6351726543795385e-05,
+      "loss": 0.5754,
+      "step": 8742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635086089585601e-05,
+      "loss": 0.5727,
+      "step": 8743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6349995168149468e-05,
+      "loss": 0.5513,
+      "step": 8744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634912936068663e-05,
+      "loss": 0.5129,
+      "step": 8745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348263473478373e-05,
+      "loss": 0.551,
+      "step": 8746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634739750653557e-05,
+      "loss": 0.5697,
+      "step": 8747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63465314598691e-05,
+      "loss": 0.5334,
+      "step": 8748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345665333489837e-05,
+      "loss": 0.5725,
+      "step": 8749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6344799127408663e-05,
+      "loss": 0.5341,
+      "step": 8750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.5669,
+      "step": 8751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343066476184098e-05,
+      "loss": 0.5721,
+      "step": 8752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634220003106247e-05,
+      "loss": 0.5671,
+      "step": 8753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341333506282454e-05,
+      "loss": 0.4849,
+      "step": 8754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340466901854933e-05,
+      "loss": 0.5515,
+      "step": 8755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6339600217790797e-05,
+      "loss": 0.5746,
+      "step": 8756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6338733454100923e-05,
+      "loss": 0.5744,
+      "step": 8757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337866610796202e-05,
+      "loss": 0.5586,
+      "step": 8758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336999687887525e-05,
+      "loss": 0.5907,
+      "step": 8759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633613268538577e-05,
+      "loss": 0.4888,
+      "step": 8760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633526560330184e-05,
+      "loss": 0.54,
+      "step": 8761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633439844164662e-05,
+      "loss": 0.5069,
+      "step": 8762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333531200430998e-05,
+      "loss": 0.5812,
+      "step": 8763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633266387966587e-05,
+      "loss": 0.5253,
+      "step": 8764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331796479362126e-05,
+      "loss": 0.5898,
+      "step": 8765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633092899953067e-05,
+      "loss": 0.5547,
+      "step": 8766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6330061440182384e-05,
+      "loss": 0.5277,
+      "step": 8767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6329193801328175e-05,
+      "loss": 0.5253,
+      "step": 8768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6328326082978938e-05,
+      "loss": 0.6364,
+      "step": 8769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632745828514557e-05,
+      "loss": 0.5624,
+      "step": 8770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632659040783897e-05,
+      "loss": 0.5964,
+      "step": 8771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6325722451070044e-05,
+      "loss": 0.5703,
+      "step": 8772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324854414849688e-05,
+      "loss": 0.6035,
+      "step": 8773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323986299188805e-05,
+      "loss": 0.6087,
+      "step": 8774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63231181040983e-05,
+      "loss": 0.5611,
+      "step": 8775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632224982958908e-05,
+      "loss": 0.5353,
+      "step": 8776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6321381475672047e-05,
+      "loss": 0.5035,
+      "step": 8777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320513042358108e-05,
+      "loss": 0.5385,
+      "step": 8778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6319644529658168e-05,
+      "loss": 0.5173,
+      "step": 8779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318775937583143e-05,
+      "loss": 0.523,
+      "step": 8780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317907266143936e-05,
+      "loss": 0.5697,
+      "step": 8781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317038515351464e-05,
+      "loss": 0.5766,
+      "step": 8782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631616968521663e-05,
+      "loss": 0.5701,
+      "step": 8783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6315300775750357e-05,
+      "loss": 0.576,
+      "step": 8784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6314431786963548e-05,
+      "loss": 0.5904,
+      "step": 8785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313562718867123e-05,
+      "loss": 0.5907,
+      "step": 8786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312693571471992e-05,
+      "loss": 0.5278,
+      "step": 8787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6311824344789082e-05,
+      "loss": 0.5846,
+      "step": 8788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310955038829305e-05,
+      "loss": 0.5376,
+      "step": 8789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310085653603576e-05,
+      "loss": 0.5829,
+      "step": 8790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630921618912282e-05,
+      "loss": 0.5246,
+      "step": 8791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308346645397954e-05,
+      "loss": 0.5693,
+      "step": 8792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63074770224399e-05,
+      "loss": 0.6246,
+      "step": 8793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306607320259583e-05,
+      "loss": 0.5071,
+      "step": 8794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305737538867928e-05,
+      "loss": 0.5588,
+      "step": 8795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630486767827585e-05,
+      "loss": 0.5278,
+      "step": 8796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6303997738494285e-05,
+      "loss": 0.6343,
+      "step": 8797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630312771953415e-05,
+      "loss": 0.5164,
+      "step": 8798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302257621406386e-05,
+      "loss": 0.5203,
+      "step": 8799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630138744412191e-05,
+      "loss": 0.5332,
+      "step": 8800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630051718769165e-05,
+      "loss": 0.5505,
+      "step": 8801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6299646852126544e-05,
+      "loss": 0.5968,
+      "step": 8802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6298776437437526e-05,
+      "loss": 0.4929,
+      "step": 8803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297905943635516e-05,
+      "loss": 0.5356,
+      "step": 8804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629703537073146e-05,
+      "loss": 0.5742,
+      "step": 8805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6296164718736285e-05,
+      "loss": 0.5159,
+      "step": 8806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295293987660925e-05,
+      "loss": 0.5971,
+      "step": 8807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6294423177516324e-05,
+      "loss": 0.4998,
+      "step": 8808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6293552288313418e-05,
+      "loss": 0.5661,
+      "step": 8809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6292681320063138e-05,
+      "loss": 0.5154,
+      "step": 8810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629181027277643e-05,
+      "loss": 0.6084,
+      "step": 8811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6290939146464235e-05,
+      "loss": 0.5933,
+      "step": 8812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6290067941137486e-05,
+      "loss": 0.5432,
+      "step": 8813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6289196656807134e-05,
+      "loss": 0.5919,
+      "step": 8814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628832529348412e-05,
+      "loss": 0.6415,
+      "step": 8815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287453851179388e-05,
+      "loss": 0.5071,
+      "step": 8816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286582329903885e-05,
+      "loss": 0.5068,
+      "step": 8817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285710729668553e-05,
+      "loss": 0.5521,
+      "step": 8818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284839050484343e-05,
+      "loss": 0.5594,
+      "step": 8819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283967292362205e-05,
+      "loss": 0.6429,
+      "step": 8820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283095455313084e-05,
+      "loss": 0.5321,
+      "step": 8821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282223539347932e-05,
+      "loss": 0.5859,
+      "step": 8822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62813515444777e-05,
+      "loss": 0.5585,
+      "step": 8823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280479470713343e-05,
+      "loss": 0.5693,
+      "step": 8824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627960731806581e-05,
+      "loss": 0.5573,
+      "step": 8825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278735086546062e-05,
+      "loss": 0.5645,
+      "step": 8826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277862776165043e-05,
+      "loss": 0.4951,
+      "step": 8827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627699038693372e-05,
+      "loss": 0.5732,
+      "step": 8828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276117918863047e-05,
+      "loss": 0.5831,
+      "step": 8829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275245371963978e-05,
+      "loss": 0.5467,
+      "step": 8830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274372746247476e-05,
+      "loss": 0.5271,
+      "step": 8831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6273500041724503e-05,
+      "loss": 0.5146,
+      "step": 8832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272627258406016e-05,
+      "loss": 0.6181,
+      "step": 8833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271754396302982e-05,
+      "loss": 0.4961,
+      "step": 8834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270881455426358e-05,
+      "loss": 0.4902,
+      "step": 8835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270008435787115e-05,
+      "loss": 0.4987,
+      "step": 8836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269135337396217e-05,
+      "loss": 0.5051,
+      "step": 8837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626826216026462e-05,
+      "loss": 0.545,
+      "step": 8838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267388904403305e-05,
+      "loss": 0.5183,
+      "step": 8839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266515569823233e-05,
+      "loss": 0.519,
+      "step": 8840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6265642156535375e-05,
+      "loss": 0.5304,
+      "step": 8841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264768664550703e-05,
+      "loss": 0.4328,
+      "step": 8842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263895093880183e-05,
+      "loss": 0.5841,
+      "step": 8843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263021444534787e-05,
+      "loss": 0.5037,
+      "step": 8844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262147716525495e-05,
+      "loss": 0.5994,
+      "step": 8845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626127390986328e-05,
+      "loss": 0.5523,
+      "step": 8846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260400024559105e-05,
+      "loss": 0.5459,
+      "step": 8847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625952606062396e-05,
+      "loss": 0.5731,
+      "step": 8848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258652018068818e-05,
+      "loss": 0.5886,
+      "step": 8849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6257777896904654e-05,
+      "loss": 0.503,
+      "step": 8850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256903697142453e-05,
+      "loss": 0.4772,
+      "step": 8851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256029418793185e-05,
+      "loss": 0.5426,
+      "step": 8852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255155061867842e-05,
+      "loss": 0.5172,
+      "step": 8853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254280626377398e-05,
+      "loss": 0.5735,
+      "step": 8854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625340611233284e-05,
+      "loss": 0.6223,
+      "step": 8855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625253151974515e-05,
+      "loss": 0.5421,
+      "step": 8856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251656848625317e-05,
+      "loss": 0.5383,
+      "step": 8857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250782098984322e-05,
+      "loss": 0.4512,
+      "step": 8858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249907270833155e-05,
+      "loss": 0.5122,
+      "step": 8859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249032364182806e-05,
+      "loss": 0.5747,
+      "step": 8860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248157379044253e-05,
+      "loss": 0.5472,
+      "step": 8861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6247282315428497e-05,
+      "loss": 0.561,
+      "step": 8862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246407173346526e-05,
+      "loss": 0.6517,
+      "step": 8863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245531952809328e-05,
+      "loss": 0.5008,
+      "step": 8864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62446566538279e-05,
+      "loss": 0.5393,
+      "step": 8865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243781276413237e-05,
+      "loss": 0.5757,
+      "step": 8866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242905820576327e-05,
+      "loss": 0.5376,
+      "step": 8867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242030286328175e-05,
+      "loss": 0.5512,
+      "step": 8868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241154673679774e-05,
+      "loss": 0.5851,
+      "step": 8869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240278982642113e-05,
+      "loss": 0.5759,
+      "step": 8870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239403213226207e-05,
+      "loss": 0.503,
+      "step": 8871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6238527365443042e-05,
+      "loss": 0.5757,
+      "step": 8872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237651439303624e-05,
+      "loss": 0.587,
+      "step": 8873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6236775434818958e-05,
+      "loss": 0.5273,
+      "step": 8874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623589935200004e-05,
+      "loss": 0.5878,
+      "step": 8875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623502319085788e-05,
+      "loss": 0.5358,
+      "step": 8876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623414695140348e-05,
+      "loss": 0.5332,
+      "step": 8877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623327063364784e-05,
+      "loss": 0.5489,
+      "step": 8878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232394237601975e-05,
+      "loss": 0.5377,
+      "step": 8879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623151776327689e-05,
+      "loss": 0.5363,
+      "step": 8880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623064121068359e-05,
+      "loss": 0.5181,
+      "step": 8881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622976457983309e-05,
+      "loss": 0.5965,
+      "step": 8882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62288878707364e-05,
+      "loss": 0.5298,
+      "step": 8883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228011083404527e-05,
+      "loss": 0.5839,
+      "step": 8884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227134217848487e-05,
+      "loss": 0.5608,
+      "step": 8885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622625727407929e-05,
+      "loss": 0.5645,
+      "step": 8886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225380252107953e-05,
+      "loss": 0.5239,
+      "step": 8887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6224503151945495e-05,
+      "loss": 0.5335,
+      "step": 8888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223625973602922e-05,
+      "loss": 0.5655,
+      "step": 8889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222748717091264e-05,
+      "loss": 0.6009,
+      "step": 8890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221871382421534e-05,
+      "loss": 0.5151,
+      "step": 8891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220993969604747e-05,
+      "loss": 0.5012,
+      "step": 8892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622011647865193e-05,
+      "loss": 0.6111,
+      "step": 8893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219238909574102e-05,
+      "loss": 0.5586,
+      "step": 8894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218361262382286e-05,
+      "loss": 0.578,
+      "step": 8895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62174835370875e-05,
+      "loss": 0.5441,
+      "step": 8896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216605733700776e-05,
+      "loss": 0.6254,
+      "step": 8897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215727852233135e-05,
+      "loss": 0.555,
+      "step": 8898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6214849892695602e-05,
+      "loss": 0.5453,
+      "step": 8899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621397185509921e-05,
+      "loss": 0.5768,
+      "step": 8900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621309373945498e-05,
+      "loss": 0.5706,
+      "step": 8901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212215545773945e-05,
+      "loss": 0.5702,
+      "step": 8902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621133727406714e-05,
+      "loss": 0.5104,
+      "step": 8903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210458924345583e-05,
+      "loss": 0.508,
+      "step": 8904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620958049662032e-05,
+      "loss": 0.5759,
+      "step": 8905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208701990902377e-05,
+      "loss": 0.5507,
+      "step": 8906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620782340720279e-05,
+      "loss": 0.5636,
+      "step": 8907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206944745532592e-05,
+      "loss": 0.4848,
+      "step": 8908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206066005902818e-05,
+      "loss": 0.6072,
+      "step": 8909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620518718832451e-05,
+      "loss": 0.533,
+      "step": 8910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204308292808704e-05,
+      "loss": 0.476,
+      "step": 8911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203429319366435e-05,
+      "loss": 0.5576,
+      "step": 8912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202550268008753e-05,
+      "loss": 0.5918,
+      "step": 8913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201671138746685e-05,
+      "loss": 0.5729,
+      "step": 8914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200791931591285e-05,
+      "loss": 0.591,
+      "step": 8915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199912646553588e-05,
+      "loss": 0.5744,
+      "step": 8916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199033283644645e-05,
+      "loss": 0.5425,
+      "step": 8917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619815384287549e-05,
+      "loss": 0.5665,
+      "step": 8918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6197274324257182e-05,
+      "loss": 0.5332,
+      "step": 8919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619639472780076e-05,
+      "loss": 0.4791,
+      "step": 8920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195515053517268e-05,
+      "loss": 0.5841,
+      "step": 8921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6194635301417763e-05,
+      "loss": 0.4942,
+      "step": 8922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193755471513295e-05,
+      "loss": 0.577,
+      "step": 8923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619287556381491e-05,
+      "loss": 0.5907,
+      "step": 8924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619199557833366e-05,
+      "loss": 0.5402,
+      "step": 8925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61911155150806e-05,
+      "loss": 0.5341,
+      "step": 8926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190235374066777e-05,
+      "loss": 0.5357,
+      "step": 8927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189355155303255e-05,
+      "loss": 0.5653,
+      "step": 8928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618847485880109e-05,
+      "loss": 0.5551,
+      "step": 8929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187594484571328e-05,
+      "loss": 0.5808,
+      "step": 8930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6186714032625036e-05,
+      "loss": 0.5571,
+      "step": 8931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185833502973267e-05,
+      "loss": 0.5682,
+      "step": 8932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6184952895627086e-05,
+      "loss": 0.5657,
+      "step": 8933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6184072210597544e-05,
+      "loss": 0.5352,
+      "step": 8934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183191447895715e-05,
+      "loss": 0.582,
+      "step": 8935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182310607532653e-05,
+      "loss": 0.5376,
+      "step": 8936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181429689519425e-05,
+      "loss": 0.5645,
+      "step": 8937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618054869386709e-05,
+      "loss": 0.5683,
+      "step": 8938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179667620586722e-05,
+      "loss": 0.4782,
+      "step": 8939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617878646968938e-05,
+      "loss": 0.5395,
+      "step": 8940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177905241186133e-05,
+      "loss": 0.5605,
+      "step": 8941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177023935088054e-05,
+      "loss": 0.5595,
+      "step": 8942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61761425514062e-05,
+      "loss": 0.5456,
+      "step": 8943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617526109015166e-05,
+      "loss": 0.5906,
+      "step": 8944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174379551335488e-05,
+      "loss": 0.4941,
+      "step": 8945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173497934968766e-05,
+      "loss": 0.5346,
+      "step": 8946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172616241062563e-05,
+      "loss": 0.5139,
+      "step": 8947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171734469627954e-05,
+      "loss": 0.6108,
+      "step": 8948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170852620676013e-05,
+      "loss": 0.5664,
+      "step": 8949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169970694217823e-05,
+      "loss": 0.5785,
+      "step": 8950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169088690264453e-05,
+      "loss": 0.5819,
+      "step": 8951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168206608826983e-05,
+      "loss": 0.5351,
+      "step": 8952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616732444991649e-05,
+      "loss": 0.581,
+      "step": 8953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616644221354406e-05,
+      "loss": 0.47,
+      "step": 8954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616555989972077e-05,
+      "loss": 0.5329,
+      "step": 8955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164677508457698e-05,
+      "loss": 0.4909,
+      "step": 8956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6163795039765936e-05,
+      "loss": 0.5136,
+      "step": 8957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616291249365656e-05,
+      "loss": 0.5362,
+      "step": 8958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616202987014066e-05,
+      "loss": 0.5635,
+      "step": 8959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616114716922932e-05,
+      "loss": 0.5465,
+      "step": 8960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160264390933627e-05,
+      "loss": 0.5438,
+      "step": 8961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159381535264668e-05,
+      "loss": 0.5475,
+      "step": 8962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615849860223353e-05,
+      "loss": 0.5751,
+      "step": 8963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6157615591851306e-05,
+      "loss": 0.569,
+      "step": 8964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156732504129085e-05,
+      "loss": 0.4893,
+      "step": 8965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155849339077957e-05,
+      "loss": 0.5226,
+      "step": 8966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615496609670902e-05,
+      "loss": 0.5199,
+      "step": 8967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154082777033364e-05,
+      "loss": 0.5936,
+      "step": 8968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6153199380062078e-05,
+      "loss": 0.5431,
+      "step": 8969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615231590580627e-05,
+      "loss": 0.5013,
+      "step": 8970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151432354277026e-05,
+      "loss": 0.5658,
+      "step": 8971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615054872548545e-05,
+      "loss": 0.5645,
+      "step": 8972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6149665019442635e-05,
+      "loss": 0.5922,
+      "step": 8973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6148781236159682e-05,
+      "loss": 0.5859,
+      "step": 8974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147897375647694e-05,
+      "loss": 0.5617,
+      "step": 8975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614701343791777e-05,
+      "loss": 0.5271,
+      "step": 8976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146129422981017e-05,
+      "loss": 0.512,
+      "step": 8977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614524533084853e-05,
+      "loss": 0.5529,
+      "step": 8978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614436116153142e-05,
+      "loss": 0.5588,
+      "step": 8979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143476915040785e-05,
+      "loss": 0.5116,
+      "step": 8980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142592591387736e-05,
+      "loss": 0.5961,
+      "step": 8981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6141708190583386e-05,
+      "loss": 0.6223,
+      "step": 8982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140823712638833e-05,
+      "loss": 0.5883,
+      "step": 8983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6139939157565195e-05,
+      "loss": 0.5871,
+      "step": 8984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613905452537357e-05,
+      "loss": 0.6091,
+      "step": 8985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613816981607508e-05,
+      "loss": 0.6031,
+      "step": 8986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6137285029680835e-05,
+      "loss": 0.5695,
+      "step": 8987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136400166201946e-05,
+      "loss": 0.5116,
+      "step": 8988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135515225649526e-05,
+      "loss": 0.4809,
+      "step": 8989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613463020803469e-05,
+      "loss": 0.5939,
+      "step": 8990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133745113368554e-05,
+      "loss": 0.5233,
+      "step": 8991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613285994166224e-05,
+      "loss": 0.5313,
+      "step": 8992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6131974692926856e-05,
+      "loss": 0.5951,
+      "step": 8993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613108936717353e-05,
+      "loss": 0.5426,
+      "step": 8994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613020396441338e-05,
+      "loss": 0.5481,
+      "step": 8995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612931848465752e-05,
+      "loss": 0.5468,
+      "step": 8996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128432927917083e-05,
+      "loss": 0.5522,
+      "step": 8997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612754729420318e-05,
+      "loss": 0.5745,
+      "step": 8998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6126661583526943e-05,
+      "loss": 0.4866,
+      "step": 8999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125775795899488e-05,
+      "loss": 0.5112,
+      "step": 9000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124889931331952e-05,
+      "loss": 0.5631,
+      "step": 9001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124003989835457e-05,
+      "loss": 0.5478,
+      "step": 9002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612311797142112e-05,
+      "loss": 0.5458,
+      "step": 9003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122231876100087e-05,
+      "loss": 0.6137,
+      "step": 9004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121345703883472e-05,
+      "loss": 0.5303,
+      "step": 9005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6120459454782415e-05,
+      "loss": 0.5741,
+      "step": 9006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6119573128808046e-05,
+      "loss": 0.5149,
+      "step": 9007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118686725971496e-05,
+      "loss": 0.6036,
+      "step": 9008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117800246283896e-05,
+      "loss": 0.5503,
+      "step": 9009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611691368975638e-05,
+      "loss": 0.519,
+      "step": 9010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116027056400092e-05,
+      "loss": 0.5524,
+      "step": 9011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115140346226157e-05,
+      "loss": 0.5355,
+      "step": 9012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114253559245718e-05,
+      "loss": 0.5687,
+      "step": 9013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611336669546991e-05,
+      "loss": 0.5795,
+      "step": 9014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112479754909876e-05,
+      "loss": 0.5098,
+      "step": 9015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111592737576756e-05,
+      "loss": 0.5317,
+      "step": 9016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110705643481682e-05,
+      "loss": 0.5765,
+      "step": 9017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610981847263581e-05,
+      "loss": 0.566,
+      "step": 9018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108931225050273e-05,
+      "loss": 0.4968,
+      "step": 9019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108043900736215e-05,
+      "loss": 0.5272,
+      "step": 9020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107156499704787e-05,
+      "loss": 0.5049,
+      "step": 9021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106269021967133e-05,
+      "loss": 0.5133,
+      "step": 9022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6105381467534392e-05,
+      "loss": 0.5962,
+      "step": 9023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104493836417723e-05,
+      "loss": 0.5932,
+      "step": 9024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103606128628268e-05,
+      "loss": 0.5717,
+      "step": 9025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102718344177178e-05,
+      "loss": 0.5248,
+      "step": 9026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101830483075603e-05,
+      "loss": 0.5193,
+      "step": 9027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6100942545334697e-05,
+      "loss": 0.5948,
+      "step": 9028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610005453096561e-05,
+      "loss": 0.5771,
+      "step": 9029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6099166439979498e-05,
+      "loss": 0.5613,
+      "step": 9030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098278272387515e-05,
+      "loss": 0.5439,
+      "step": 9031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609739002820081e-05,
+      "loss": 0.5328,
+      "step": 9032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6096501707430547e-05,
+      "loss": 0.4912,
+      "step": 9033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6095613310087884e-05,
+      "loss": 0.5765,
+      "step": 9034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094724836183972e-05,
+      "loss": 0.5042,
+      "step": 9035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609383628572998e-05,
+      "loss": 0.5761,
+      "step": 9036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092947658737057e-05,
+      "loss": 0.5637,
+      "step": 9037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092058955216378e-05,
+      "loss": 0.5068,
+      "step": 9038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609117017517909e-05,
+      "loss": 0.559,
+      "step": 9039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6090281318636365e-05,
+      "loss": 0.423,
+      "step": 9040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089392385599366e-05,
+      "loss": 0.5859,
+      "step": 9041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608850337607926e-05,
+      "loss": 0.5187,
+      "step": 9042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.4921,
+      "step": 9043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608672512763438e-05,
+      "loss": 0.4866,
+      "step": 9044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6085835888731946e-05,
+      "loss": 0.5942,
+      "step": 9045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6084946573391068e-05,
+      "loss": 0.5514,
+      "step": 9046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6084057181622922e-05,
+      "loss": 0.5414,
+      "step": 9047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608316771343868e-05,
+      "loss": 0.5933,
+      "step": 9048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082278168849506e-05,
+      "loss": 0.5297,
+      "step": 9049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608138854786658e-05,
+      "loss": 0.5764,
+      "step": 9050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080498850501074e-05,
+      "loss": 0.5519,
+      "step": 9051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607960907676416e-05,
+      "loss": 0.5786,
+      "step": 9052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078719226667017e-05,
+      "loss": 0.5073,
+      "step": 9053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607782930022082e-05,
+      "loss": 0.4712,
+      "step": 9054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076939297436743e-05,
+      "loss": 0.5684,
+      "step": 9055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607604921832597e-05,
+      "loss": 0.4996,
+      "step": 9056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075159062899683e-05,
+      "loss": 0.5415,
+      "step": 9057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074268831169056e-05,
+      "loss": 0.5662,
+      "step": 9058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6073378523145272e-05,
+      "loss": 0.5124,
+      "step": 9059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607248813883951e-05,
+      "loss": 0.5353,
+      "step": 9060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607159767826296e-05,
+      "loss": 0.6061,
+      "step": 9061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070707141426805e-05,
+      "loss": 0.5053,
+      "step": 9062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069816528342222e-05,
+      "loss": 0.5112,
+      "step": 9063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068925839020408e-05,
+      "loss": 0.5189,
+      "step": 9064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068035073472544e-05,
+      "loss": 0.4967,
+      "step": 9065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606714423170982e-05,
+      "loss": 0.5054,
+      "step": 9066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066253313743427e-05,
+      "loss": 0.558,
+      "step": 9067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065362319584548e-05,
+      "loss": 0.4889,
+      "step": 9068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064471249244384e-05,
+      "loss": 0.6319,
+      "step": 9069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063580102734117e-05,
+      "loss": 0.6035,
+      "step": 9070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062688880064947e-05,
+      "loss": 0.5418,
+      "step": 9071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061797581248067e-05,
+      "loss": 0.5229,
+      "step": 9072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6060906206294665e-05,
+      "loss": 0.4657,
+      "step": 9073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606001475521594e-05,
+      "loss": 0.5165,
+      "step": 9074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60591232280231e-05,
+      "loss": 0.5465,
+      "step": 9075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6058231624727324e-05,
+      "loss": 0.5705,
+      "step": 9076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605733994533982e-05,
+      "loss": 0.6604,
+      "step": 9077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056448189871795e-05,
+      "loss": 0.6178,
+      "step": 9078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055556358334438e-05,
+      "loss": 0.5427,
+      "step": 9079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054664450738946e-05,
+      "loss": 0.58,
+      "step": 9080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6053772467096534e-05,
+      "loss": 0.4804,
+      "step": 9081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6052880407418404e-05,
+      "loss": 0.5753,
+      "step": 9082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051988271715754e-05,
+      "loss": 0.5237,
+      "step": 9083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051096059999798e-05,
+      "loss": 0.4951,
+      "step": 9084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050203772281733e-05,
+      "loss": 0.546,
+      "step": 9085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6049311408572765e-05,
+      "loss": 0.5448,
+      "step": 9086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604841896888411e-05,
+      "loss": 0.4771,
+      "step": 9087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047526453226977e-05,
+      "loss": 0.6295,
+      "step": 9088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604663386161257e-05,
+      "loss": 0.5746,
+      "step": 9089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045741194052107e-05,
+      "loss": 0.5434,
+      "step": 9090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044848450556788e-05,
+      "loss": 0.5762,
+      "step": 9091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043955631137843e-05,
+      "loss": 0.4798,
+      "step": 9092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604306273580647e-05,
+      "loss": 0.5739,
+      "step": 9093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6042169764573892e-05,
+      "loss": 0.5251,
+      "step": 9094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6041276717451326e-05,
+      "loss": 0.4788,
+      "step": 9095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040383594449986e-05,
+      "loss": 0.5206,
+      "step": 9096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039490395581087e-05,
+      "loss": 0.5195,
+      "step": 9097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603859712085585e-05,
+      "loss": 0.4653,
+      "step": 9098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60377037702855e-05,
+      "loss": 0.5399,
+      "step": 9099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036810343881247e-05,
+      "loss": 0.513,
+      "step": 9100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035916841654322e-05,
+      "loss": 0.5908,
+      "step": 9101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603502326361594e-05,
+      "loss": 0.569,
+      "step": 9102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603412960977733e-05,
+      "loss": 0.5373,
+      "step": 9103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6033235880149716e-05,
+      "loss": 0.5208,
+      "step": 9104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603234207474432e-05,
+      "loss": 0.5243,
+      "step": 9105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6031448193572367e-05,
+      "loss": 0.6468,
+      "step": 9106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6030554236645087e-05,
+      "loss": 0.5621,
+      "step": 9107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602966020397371e-05,
+      "loss": 0.4897,
+      "step": 9108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028766095569462e-05,
+      "loss": 0.5143,
+      "step": 9109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027871911443576e-05,
+      "loss": 0.6194,
+      "step": 9110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602697765160728e-05,
+      "loss": 0.5208,
+      "step": 9111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026083316071807e-05,
+      "loss": 0.5648,
+      "step": 9112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602518890484839e-05,
+      "loss": 0.5674,
+      "step": 9113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024294417948264e-05,
+      "loss": 0.5913,
+      "step": 9114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023399855382663e-05,
+      "loss": 0.6018,
+      "step": 9115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6022505217162824e-05,
+      "loss": 0.5036,
+      "step": 9116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021610503299977e-05,
+      "loss": 0.5818,
+      "step": 9117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602071571380537e-05,
+      "loss": 0.5326,
+      "step": 9118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601982084869023e-05,
+      "loss": 0.5363,
+      "step": 9119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601892590796581e-05,
+      "loss": 0.5941,
+      "step": 9120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601803089164334e-05,
+      "loss": 0.518,
+      "step": 9121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017135799734064e-05,
+      "loss": 0.597,
+      "step": 9122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016240632249224e-05,
+      "loss": 0.5014,
+      "step": 9123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601534538920007e-05,
+      "loss": 0.5326,
+      "step": 9124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601445007059784e-05,
+      "loss": 0.6263,
+      "step": 9125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601355467645378e-05,
+      "loss": 0.5668,
+      "step": 9126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012659206779134e-05,
+      "loss": 0.5296,
+      "step": 9127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601176366158515e-05,
+      "loss": 0.5726,
+      "step": 9128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601086804088308e-05,
+      "loss": 0.5767,
+      "step": 9129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009972344684172e-05,
+      "loss": 0.5484,
+      "step": 9130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009076572999676e-05,
+      "loss": 0.5497,
+      "step": 9131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600818072584084e-05,
+      "loss": 0.5108,
+      "step": 9132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6007284803218916e-05,
+      "loss": 0.5146,
+      "step": 9133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006388805145163e-05,
+      "loss": 0.5889,
+      "step": 9134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6005492731630826e-05,
+      "loss": 0.5813,
+      "step": 9135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004596582687163e-05,
+      "loss": 0.5147,
+      "step": 9136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600370035832543e-05,
+      "loss": 0.5329,
+      "step": 9137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002804058556888e-05,
+      "loss": 0.5496,
+      "step": 9138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600190768339279e-05,
+      "loss": 0.5228,
+      "step": 9139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6001011232844395e-05,
+      "loss": 0.4712,
+      "step": 9140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000114706922962e-05,
+      "loss": 0.6075,
+      "step": 9141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999218105639757e-05,
+      "loss": 0.5035,
+      "step": 9142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599832142900603e-05,
+      "loss": 0.533,
+      "step": 9143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997424677033055e-05,
+      "loss": 0.5463,
+      "step": 9144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5996527849732088e-05,
+      "loss": 0.4823,
+      "step": 9145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599563094711439e-05,
+      "loss": 0.6205,
+      "step": 9146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599473396919124e-05,
+      "loss": 0.5839,
+      "step": 9147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599383691597389e-05,
+      "loss": 0.5528,
+      "step": 9148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992939787473616e-05,
+      "loss": 0.4901,
+      "step": 9149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992042583701683e-05,
+      "loss": 0.5587,
+      "step": 9150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5991145304669355e-05,
+      "loss": 0.5147,
+      "step": 9151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599024795038791e-05,
+      "loss": 0.6663,
+      "step": 9152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989350520868616e-05,
+      "loss": 0.5532,
+      "step": 9153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988453016122743e-05,
+      "loss": 0.5518,
+      "step": 9154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5987555436161564e-05,
+      "loss": 0.5691,
+      "step": 9155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5986657780996354e-05,
+      "loss": 0.5306,
+      "step": 9156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985760050638387e-05,
+      "loss": 0.5938,
+      "step": 9157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5984862245098937e-05,
+      "loss": 0.655,
+      "step": 9158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983964364389284e-05,
+      "loss": 0.4963,
+      "step": 9159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983066408520704e-05,
+      "loss": 0.55,
+      "step": 9160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598216837750447e-05,
+      "loss": 0.5809,
+      "step": 9161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598127027135187e-05,
+      "loss": 0.5478,
+      "step": 9162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5980372090074182e-05,
+      "loss": 0.5905,
+      "step": 9163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5979473833682683e-05,
+      "loss": 0.538,
+      "step": 9164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597857550218866e-05,
+      "loss": 0.5486,
+      "step": 9165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597767709560339e-05,
+      "loss": 0.5133,
+      "step": 9166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976778613938163e-05,
+      "loss": 0.6258,
+      "step": 9167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975880057204264e-05,
+      "loss": 0.5749,
+      "step": 9168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974981425412974e-05,
+      "loss": 0.4544,
+      "step": 9169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974082718575584e-05,
+      "loss": 0.5596,
+      "step": 9170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597318393670338e-05,
+      "loss": 0.5737,
+      "step": 9171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972285079807653e-05,
+      "loss": 0.559,
+      "step": 9172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597138614789969e-05,
+      "loss": 0.5408,
+      "step": 9173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597048714099078e-05,
+      "loss": 0.5492,
+      "step": 9174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596958805909222e-05,
+      "loss": 0.6516,
+      "step": 9175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5968688902215296e-05,
+      "loss": 0.5,
+      "step": 9176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596778967037131e-05,
+      "loss": 0.5212,
+      "step": 9177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596689036357155e-05,
+      "loss": 0.5609,
+      "step": 9178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596599098182731e-05,
+      "loss": 0.4732,
+      "step": 9179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596509152514989e-05,
+      "loss": 0.5743,
+      "step": 9180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596419199355059e-05,
+      "loss": 0.5724,
+      "step": 9181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59632923870407e-05,
+      "loss": 0.5184,
+      "step": 9182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962392705631527e-05,
+      "loss": 0.5736,
+      "step": 9183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961492949334366e-05,
+      "loss": 0.5315,
+      "step": 9184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596059311816052e-05,
+      "loss": 0.5345,
+      "step": 9185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595969321212129e-05,
+      "loss": 0.5729,
+      "step": 9186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595879323122798e-05,
+      "loss": 0.5466,
+      "step": 9187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595789317549189e-05,
+      "loss": 0.4715,
+      "step": 9188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956993044924334e-05,
+      "loss": 0.5235,
+      "step": 9189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956092839536614e-05,
+      "loss": 0.5145,
+      "step": 9190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955192559340024e-05,
+      "loss": 0.5148,
+      "step": 9191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954292204345888e-05,
+      "loss": 0.5276,
+      "step": 9192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595339177456551e-05,
+      "loss": 0.4984,
+      "step": 9193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952491270010196e-05,
+      "loss": 0.5852,
+      "step": 9194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5951590690691256e-05,
+      "loss": 0.5962,
+      "step": 9195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950690036620008e-05,
+      "loss": 0.5112,
+      "step": 9196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949789307807756e-05,
+      "loss": 0.5368,
+      "step": 9197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5948888504265818e-05,
+      "loss": 0.5206,
+      "step": 9198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594798762600551e-05,
+      "loss": 0.581,
+      "step": 9199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594708667303814e-05,
+      "loss": 0.5195,
+      "step": 9200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594618564537503e-05,
+      "loss": 0.5874,
+      "step": 9201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5945284543027497e-05,
+      "loss": 0.5366,
+      "step": 9202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944383366006856e-05,
+      "loss": 0.4936,
+      "step": 9203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5943482114324425e-05,
+      "loss": 0.531,
+      "step": 9204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942580787991526e-05,
+      "loss": 0.5668,
+      "step": 9205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941679387019484e-05,
+      "loss": 0.5688,
+      "step": 9206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940777911419613e-05,
+      "loss": 0.5216,
+      "step": 9207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593987636120324e-05,
+      "loss": 0.5406,
+      "step": 9208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938974736381686e-05,
+      "loss": 0.6084,
+      "step": 9209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938073036966275e-05,
+      "loss": 0.5177,
+      "step": 9210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937171262968335e-05,
+      "loss": 0.5174,
+      "step": 9211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593626941439919e-05,
+      "loss": 0.5024,
+      "step": 9212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593536749127017e-05,
+      "loss": 0.5043,
+      "step": 9213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5934465493592606e-05,
+      "loss": 0.5547,
+      "step": 9214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933563421377822e-05,
+      "loss": 0.5328,
+      "step": 9215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5932661274637146e-05,
+      "loss": 0.4993,
+      "step": 9216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931759053381914e-05,
+      "loss": 0.5678,
+      "step": 9217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5930856757623453e-05,
+      "loss": 0.6407,
+      "step": 9218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929954387373103e-05,
+      "loss": 0.6609,
+      "step": 9219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592905194264219e-05,
+      "loss": 0.5456,
+      "step": 9220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5928149423442055e-05,
+      "loss": 0.5337,
+      "step": 9221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592724682978403e-05,
+      "loss": 0.5932,
+      "step": 9222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5926344161679454e-05,
+      "loss": 0.5377,
+      "step": 9223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925441419139665e-05,
+      "loss": 0.5516,
+      "step": 9224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5924538602175997e-05,
+      "loss": 0.551,
+      "step": 9225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923635710799794e-05,
+      "loss": 0.5638,
+      "step": 9226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5922732745022394e-05,
+      "loss": 0.5539,
+      "step": 9227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5921829704855136e-05,
+      "loss": 0.4886,
+      "step": 9228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592092659030937e-05,
+      "loss": 0.5007,
+      "step": 9229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920023401396435e-05,
+      "loss": 0.4971,
+      "step": 9230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591912013812767e-05,
+      "loss": 0.5808,
+      "step": 9231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5918216800514426e-05,
+      "loss": 0.5021,
+      "step": 9232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591731338856805e-05,
+      "loss": 0.564,
+      "step": 9233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916409902299882e-05,
+      "loss": 0.604,
+      "step": 9234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591550634172128e-05,
+      "loss": 0.5791,
+      "step": 9235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914602706843584e-05,
+      "loss": 0.5494,
+      "step": 9236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5913698997678144e-05,
+      "loss": 0.5734,
+      "step": 9237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912795214236317e-05,
+      "loss": 0.4547,
+      "step": 9238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591189135652945e-05,
+      "loss": 0.5739,
+      "step": 9239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910987424568897e-05,
+      "loss": 0.532,
+      "step": 9240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591008341836601e-05,
+      "loss": 0.6201,
+      "step": 9241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909179337932145e-05,
+      "loss": 0.5979,
+      "step": 9242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5908275183278658e-05,
+      "loss": 0.5737,
+      "step": 9243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907370954416902e-05,
+      "loss": 0.5583,
+      "step": 9244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906466651358234e-05,
+      "loss": 0.6224,
+      "step": 9245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905562274114016e-05,
+      "loss": 0.5901,
+      "step": 9246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904657822695606e-05,
+      "loss": 0.5615,
+      "step": 9247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590375329711436e-05,
+      "loss": 0.4495,
+      "step": 9248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5902848697381647e-05,
+      "loss": 0.5766,
+      "step": 9249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901944023508818e-05,
+      "loss": 0.5179,
+      "step": 9250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901039275507247e-05,
+      "loss": 0.5153,
+      "step": 9251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590013445338829e-05,
+      "loss": 0.5053,
+      "step": 9252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899229557163312e-05,
+      "loss": 0.5842,
+      "step": 9253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898324586843683e-05,
+      "loss": 0.5912,
+      "step": 9254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5897419542440767e-05,
+      "loss": 0.546,
+      "step": 9255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896514423965927e-05,
+      "loss": 0.5104,
+      "step": 9256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589560923143054e-05,
+      "loss": 0.6034,
+      "step": 9257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589470396484597e-05,
+      "loss": 0.542,
+      "step": 9258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5893798624223585e-05,
+      "loss": 0.6028,
+      "step": 9259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892893209574765e-05,
+      "loss": 0.5591,
+      "step": 9260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891987720910874e-05,
+      "loss": 0.6429,
+      "step": 9261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891082158243283e-05,
+      "loss": 0.5299,
+      "step": 9262
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5890176521583373e-05,
+      "loss": 0.5681,
+      "step": 9263
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588927081094252e-05,
+      "loss": 0.5946,
+      "step": 9264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888365026332092e-05,
+      "loss": 0.5751,
+      "step": 9265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588745916776347e-05,
+      "loss": 0.537,
+      "step": 9266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588655323524803e-05,
+      "loss": 0.5522,
+      "step": 9267
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588564722879715e-05,
+      "loss": 0.5872,
+      "step": 9268
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884741148422215e-05,
+      "loss": 0.4818,
+      "step": 9269
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58838349941346e-05,
+      "loss": 0.5705,
+      "step": 9270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882928765945687e-05,
+      "loss": 0.5358,
+      "step": 9271
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588202246386686e-05,
+      "loss": 0.533,
+      "step": 9272
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881116087909504e-05,
+      "loss": 0.5051,
+      "step": 9273
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880209638084995e-05,
+      "loss": 0.622,
+      "step": 9274
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5879303114404726e-05,
+      "loss": 0.5574,
+      "step": 9275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878396516880083e-05,
+      "loss": 0.597,
+      "step": 9276
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587748984552245e-05,
+      "loss": 0.5539,
+      "step": 9277
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5876583100343213e-05,
+      "loss": 0.559,
+      "step": 9278
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875676281353766e-05,
+      "loss": 0.4862,
+      "step": 9279
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874769388565493e-05,
+      "loss": 0.6238,
+      "step": 9280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873862421989788e-05,
+      "loss": 0.5123,
+      "step": 9281
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872955381638045e-05,
+      "loss": 0.5671,
+      "step": 9282
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587204826752165e-05,
+      "loss": 0.5738,
+      "step": 9283
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5871141079652e-05,
+      "loss": 0.5255,
+      "step": 9284
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5870233818040493e-05,
+      "loss": 0.5342,
+      "step": 9285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586932648269852e-05,
+      "loss": 0.528,
+      "step": 9286
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868419073637474e-05,
+      "loss": 0.5541,
+      "step": 9287
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5867511590868754e-05,
+      "loss": 0.5808,
+      "step": 9288
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5866604034403767e-05,
+      "loss": 0.4811,
+      "step": 9289
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865696404253896e-05,
+      "loss": 0.6038,
+      "step": 9290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864788700430556e-05,
+      "loss": 0.5044,
+      "step": 9291
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863880922945142e-05,
+      "loss": 0.5524,
+      "step": 9292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586297307180905e-05,
+      "loss": 0.584,
+      "step": 9293
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586206514703369e-05,
+      "loss": 0.5705,
+      "step": 9294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586115714863047e-05,
+      "loss": 0.5802,
+      "step": 9295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586024907661078e-05,
+      "loss": 0.5268,
+      "step": 9296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859340930986037e-05,
+      "loss": 0.6008,
+      "step": 9297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585843271176764e-05,
+      "loss": 0.5233,
+      "step": 9298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5857524418967003e-05,
+      "loss": 0.5659,
+      "step": 9299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856616052595532e-05,
+      "loss": 0.5526,
+      "step": 9300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855707612664633e-05,
+      "loss": 0.5657,
+      "step": 9301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5854799099185722e-05,
+      "loss": 0.4382,
+      "step": 9302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58538905121702e-05,
+      "loss": 0.5859,
+      "step": 9303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5852981851629494e-05,
+      "loss": 0.5701,
+      "step": 9304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5852073117575003e-05,
+      "loss": 0.5504,
+      "step": 9305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5851164310018145e-05,
+      "loss": 0.5061,
+      "step": 9306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850255428970336e-05,
+      "loss": 0.5764,
+      "step": 9307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584934647444299e-05,
+      "loss": 0.504,
+      "step": 9308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584843744644753e-05,
+      "loss": 0.557,
+      "step": 9309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847528344995364e-05,
+      "loss": 0.5699,
+      "step": 9310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5846619170097914e-05,
+      "loss": 0.5637,
+      "step": 9311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5845709921766604e-05,
+      "loss": 0.5031,
+      "step": 9312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5844800600012845e-05,
+      "loss": 0.5292,
+      "step": 9313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843891204848065e-05,
+      "loss": 0.5308,
+      "step": 9314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842981736283686e-05,
+      "loss": 0.4141,
+      "step": 9315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584207219433113e-05,
+      "loss": 0.5413,
+      "step": 9316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841162579001815e-05,
+      "loss": 0.5462,
+      "step": 9317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5840252890307173e-05,
+      "loss": 0.5448,
+      "step": 9318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5839343128258627e-05,
+      "loss": 0.5193,
+      "step": 9319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838433292867606e-05,
+      "loss": 0.6171,
+      "step": 9320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583752338414554e-05,
+      "loss": 0.5529,
+      "step": 9321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836613402103846e-05,
+      "loss": 0.5222,
+      "step": 9322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5835703346753963e-05,
+      "loss": 0.5293,
+      "step": 9323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5834793218107322e-05,
+      "loss": 0.5279,
+      "step": 9324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5833883016175345e-05,
+      "loss": 0.5766,
+      "step": 9325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5832972740969473e-05,
+      "loss": 0.5712,
+      "step": 9326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583206239250114e-05,
+      "loss": 0.5894,
+      "step": 9327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831151970781775e-05,
+      "loss": 0.5014,
+      "step": 9328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5830241475822817e-05,
+      "loss": 0.4407,
+      "step": 9329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829330907635695e-05,
+      "loss": 0.5766,
+      "step": 9330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5828420266231854e-05,
+      "loss": 0.5857,
+      "step": 9331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827509551622724e-05,
+      "loss": 0.5378,
+      "step": 9332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826598763819753e-05,
+      "loss": 0.5501,
+      "step": 9333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582568790283437e-05,
+      "loss": 0.5372,
+      "step": 9334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824776968678024e-05,
+      "loss": 0.5755,
+      "step": 9335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582386596136215e-05,
+      "loss": 0.6028,
+      "step": 9336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822954880898197e-05,
+      "loss": 0.5381,
+      "step": 9337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822043727297603e-05,
+      "loss": 0.5427,
+      "step": 9338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821132500571814e-05,
+      "loss": 0.605,
+      "step": 9339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820221200732272e-05,
+      "loss": 0.5295,
+      "step": 9340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581930982779043e-05,
+      "loss": 0.5167,
+      "step": 9341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818398381757728e-05,
+      "loss": 0.5241,
+      "step": 9342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581748686264562e-05,
+      "loss": 0.5627,
+      "step": 9343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816575270465547e-05,
+      "loss": 0.5393,
+      "step": 9344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5815663605228965e-05,
+      "loss": 0.584,
+      "step": 9345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581475186694732e-05,
+      "loss": 0.5432,
+      "step": 9346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813840055632067e-05,
+      "loss": 0.6236,
+      "step": 9347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581292817129466e-05,
+      "loss": 0.5362,
+      "step": 9348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812016213946547e-05,
+      "loss": 0.6166,
+      "step": 9349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5811104183599185e-05,
+      "loss": 0.5251,
+      "step": 9350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810192080264027e-05,
+      "loss": 0.5885,
+      "step": 9351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809279903952535e-05,
+      "loss": 0.5702,
+      "step": 9352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580836765467616e-05,
+      "loss": 0.5766,
+      "step": 9353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5807455332446365e-05,
+      "loss": 0.5594,
+      "step": 9354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5806542937274604e-05,
+      "loss": 0.5683,
+      "step": 9355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5805630469172334e-05,
+      "loss": 0.5595,
+      "step": 9356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804717928151024e-05,
+      "loss": 0.5821,
+      "step": 9357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5803805314222135e-05,
+      "loss": 0.5684,
+      "step": 9358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580289262739712e-05,
+      "loss": 0.5493,
+      "step": 9359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580197986768745e-05,
+      "loss": 0.5326,
+      "step": 9360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801067035104593e-05,
+      "loss": 0.6174,
+      "step": 9361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580015412966e-05,
+      "loss": 0.5275,
+      "step": 9362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799241151365155e-05,
+      "loss": 0.5633,
+      "step": 9363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579832810023151e-05,
+      "loss": 0.5521,
+      "step": 9364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797414976270545e-05,
+      "loss": 0.5648,
+      "step": 9365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579650177949372e-05,
+      "loss": 0.5667,
+      "step": 9366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579558850991251e-05,
+      "loss": 0.58,
+      "step": 9367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794675167538382e-05,
+      "loss": 0.5335,
+      "step": 9368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579376175238281e-05,
+      "loss": 0.5456,
+      "step": 9369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5792848264457263e-05,
+      "loss": 0.5616,
+      "step": 9370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579193470377322e-05,
+      "loss": 0.4842,
+      "step": 9371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791021070342155e-05,
+      "loss": 0.5669,
+      "step": 9372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790107364175536e-05,
+      "loss": 0.6124,
+      "step": 9373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789193585284847e-05,
+      "loss": 0.5538,
+      "step": 9374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5788279733681564e-05,
+      "loss": 0.6418,
+      "step": 9375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578736580937716e-05,
+      "loss": 0.5519,
+      "step": 9376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578645181238312e-05,
+      "loss": 0.4712,
+      "step": 9377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578553774271092e-05,
+      "loss": 0.5183,
+      "step": 9378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578462360037204e-05,
+      "loss": 0.5559,
+      "step": 9379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578370938537797e-05,
+      "loss": 0.5993,
+      "step": 9380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578279509774018e-05,
+      "loss": 0.511,
+      "step": 9381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578188073747016e-05,
+      "loss": 0.5455,
+      "step": 9382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780966304579398e-05,
+      "loss": 0.5177,
+      "step": 9383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780051799079374e-05,
+      "loss": 0.5525,
+      "step": 9384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779137220981572e-05,
+      "loss": 0.5767,
+      "step": 9385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778222570297486e-05,
+      "loss": 0.494,
+      "step": 9386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57773078470386e-05,
+      "loss": 0.5788,
+      "step": 9387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776393051216404e-05,
+      "loss": 0.5864,
+      "step": 9388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577547818284239e-05,
+      "loss": 0.557,
+      "step": 9389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5774563241928045e-05,
+      "loss": 0.5151,
+      "step": 9390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5773648228484863e-05,
+      "loss": 0.5484,
+      "step": 9391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5772733142524335e-05,
+      "loss": 0.4911,
+      "step": 9392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5771817984057952e-05,
+      "loss": 0.6362,
+      "step": 9393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5770902753097218e-05,
+      "loss": 0.5434,
+      "step": 9394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769987449653618e-05,
+      "loss": 0.5377,
+      "step": 9395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769072073738654e-05,
+      "loss": 0.5829,
+      "step": 9396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576815662536382e-05,
+      "loss": 0.5764,
+      "step": 9397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576724110454062e-05,
+      "loss": 0.6175,
+      "step": 9398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5766325511280546e-05,
+      "loss": 0.4914,
+      "step": 9399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57654098455951e-05,
+      "loss": 0.5186,
+      "step": 9400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5764494107495784e-05,
+      "loss": 0.5833,
+      "step": 9401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57635782969941e-05,
+      "loss": 0.4584,
+      "step": 9402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762662414101547e-05,
+      "loss": 0.539,
+      "step": 9403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576174645882963e-05,
+      "loss": 0.5711,
+      "step": 9404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5760830431189862e-05,
+      "loss": 0.5499,
+      "step": 9405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759914331193738e-05,
+      "loss": 0.5417,
+      "step": 9406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758998158852766e-05,
+      "loss": 0.5332,
+      "step": 9407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758081914178457e-05,
+      "loss": 0.5695,
+      "step": 9408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757165597182314e-05,
+      "loss": 0.5983,
+      "step": 9409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575624920787585e-05,
+      "loss": 0.5167,
+      "step": 9410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755332746270574e-05,
+      "loss": 0.5767,
+      "step": 9411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575441621237799e-05,
+      "loss": 0.6029,
+      "step": 9412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753499606209623e-05,
+      "loss": 0.5179,
+      "step": 9413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575258292777698e-05,
+      "loss": 0.5677,
+      "step": 9414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5751666177091568e-05,
+      "loss": 0.5376,
+      "step": 9415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750749354164908e-05,
+      "loss": 0.4954,
+      "step": 9416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574983245900852e-05,
+      "loss": 0.4685,
+      "step": 9417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5748915491633906e-05,
+      "loss": 0.521,
+      "step": 9418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747998452052594e-05,
+      "loss": 0.5428,
+      "step": 9419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57470813402761e-05,
+      "loss": 0.5647,
+      "step": 9420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746164156315943e-05,
+      "loss": 0.6049,
+      "step": 9421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745246900183647e-05,
+      "loss": 0.5131,
+      "step": 9422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574432957189072e-05,
+      "loss": 0.5352,
+      "step": 9423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743412171448696e-05,
+      "loss": 0.5229,
+      "step": 9424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742494698869094e-05,
+      "loss": 0.5639,
+      "step": 9425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741577154163438e-05,
+      "loss": 0.5836,
+      "step": 9426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5740659537343246e-05,
+      "loss": 0.4861,
+      "step": 9427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5739741848420052e-05,
+      "loss": 0.5763,
+      "step": 9428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573882408740538e-05,
+      "loss": 0.5024,
+      "step": 9429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737906254310753e-05,
+      "loss": 0.5624,
+      "step": 9430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736988349147705e-05,
+      "loss": 0.5455,
+      "step": 9431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573607037192776e-05,
+      "loss": 0.4832,
+      "step": 9432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573515232266245e-05,
+      "loss": 0.5366,
+      "step": 9433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573423420136331e-05,
+      "loss": 0.5668,
+      "step": 9434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733316008041864e-05,
+      "loss": 0.5518,
+      "step": 9435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732397742709643e-05,
+      "loss": 0.5535,
+      "step": 9436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731479405378194e-05,
+      "loss": 0.5088,
+      "step": 9437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730560996059035e-05,
+      "loss": 0.5249,
+      "step": 9438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5729642514763715e-05,
+      "loss": 0.5488,
+      "step": 9439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572872396150376e-05,
+      "loss": 0.5561,
+      "step": 9440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727805336290714e-05,
+      "loss": 0.5743,
+      "step": 9441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726886639136112e-05,
+      "loss": 0.5659,
+      "step": 9442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572596787005149e-05,
+      "loss": 0.6266,
+      "step": 9443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5725049029048395e-05,
+      "loss": 0.5078,
+      "step": 9444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572413011613836e-05,
+      "loss": 0.5995,
+      "step": 9445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723211131332932e-05,
+      "loss": 0.5759,
+      "step": 9446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722292074643652e-05,
+      "loss": 0.493,
+      "step": 9447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572137294608206e-05,
+      "loss": 0.5552,
+      "step": 9448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572045374565971e-05,
+      "loss": 0.5275,
+      "step": 9449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5719534473388134e-05,
+      "loss": 0.5117,
+      "step": 9450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718615129278888e-05,
+      "loss": 0.5514,
+      "step": 9451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571769571334351e-05,
+      "loss": 0.5909,
+      "step": 9452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571677622559356e-05,
+      "loss": 0.4687,
+      "step": 9453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571585666604058e-05,
+      "loss": 0.4993,
+      "step": 9454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571493703469612e-05,
+      "loss": 0.5817,
+      "step": 9455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571401733157173e-05,
+      "loss": 0.4809,
+      "step": 9456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713097556678963e-05,
+      "loss": 0.4879,
+      "step": 9457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712177710029365e-05,
+      "loss": 0.6092,
+      "step": 9458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57112577916345e-05,
+      "loss": 0.5617,
+      "step": 9459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5710337801505918e-05,
+      "loss": 0.6002,
+      "step": 9460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5709417739655176e-05,
+      "loss": 0.5664,
+      "step": 9461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570849760609382e-05,
+      "loss": 0.5666,
+      "step": 9462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570757740083342e-05,
+      "loss": 0.5744,
+      "step": 9463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706657123885527e-05,
+      "loss": 0.542,
+      "step": 9464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57057367752617e-05,
+      "loss": 0.5946,
+      "step": 9465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57048163549735e-05,
+      "loss": 0.5902,
+      "step": 9466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5703895863032485e-05,
+      "loss": 0.5756,
+      "step": 9467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702975299450223e-05,
+      "loss": 0.5597,
+      "step": 9468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702054664238268e-05,
+      "loss": 0.5667,
+      "step": 9469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570113395740819e-05,
+      "loss": 0.578,
+      "step": 9470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5700213178971548e-05,
+      "loss": 0.5246,
+      "step": 9471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699292328939907e-05,
+      "loss": 0.5181,
+      "step": 9472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569837140732484e-05,
+      "loss": 0.6056,
+      "step": 9473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697450414137908e-05,
+      "loss": 0.5525,
+      "step": 9474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696529349390676e-05,
+      "loss": 0.5575,
+      "step": 9475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569560821309472e-05,
+      "loss": 0.5059,
+      "step": 9476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694687005261604e-05,
+      "loss": 0.5439,
+      "step": 9477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5693765725902902e-05,
+      "loss": 0.5502,
+      "step": 9478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569284437503018e-05,
+      "loss": 0.4932,
+      "step": 9479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569192295265502e-05,
+      "loss": 0.5689,
+      "step": 9480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691001458788984e-05,
+      "loss": 0.5687,
+      "step": 9481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569007989344365e-05,
+      "loss": 0.6269,
+      "step": 9482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5689158256630597e-05,
+      "loss": 0.5359,
+      "step": 9483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688236548361397e-05,
+      "loss": 0.5754,
+      "step": 9484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568731476864763e-05,
+      "loss": 0.6061,
+      "step": 9485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5686392917500865e-05,
+      "loss": 0.4673,
+      "step": 9486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568547099493269e-05,
+      "loss": 0.6629,
+      "step": 9487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568454900095468e-05,
+      "loss": 0.5975,
+      "step": 9488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568362693557842e-05,
+      "loss": 0.5528,
+      "step": 9489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5682704798815484e-05,
+      "loss": 0.5154,
+      "step": 9490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5681782590677457e-05,
+      "loss": 0.4924,
+      "step": 9491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5680860311175922e-05,
+      "loss": 0.5778,
+      "step": 9492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679937960322466e-05,
+      "loss": 0.5746,
+      "step": 9493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567901553812867e-05,
+      "loss": 0.5089,
+      "step": 9494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5678093044606116e-05,
+      "loss": 0.6437,
+      "step": 9495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56771704797664e-05,
+      "loss": 0.5016,
+      "step": 9496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676247843621105e-05,
+      "loss": 0.5727,
+      "step": 9497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5675325136181818e-05,
+      "loss": 0.5427,
+      "step": 9498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567440235746013e-05,
+      "loss": 0.4895,
+      "step": 9499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673479507467634e-05,
+      "loss": 0.5304,
+      "step": 9500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5672556586215913e-05,
+      "loss": 0.5294,
+      "step": 9501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671633593716565e-05,
+      "loss": 0.5262,
+      "step": 9502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5670710529981183e-05,
+      "loss": 0.5812,
+      "step": 9503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669787395021355e-05,
+      "loss": 0.5891,
+      "step": 9504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5668864188848682e-05,
+      "loss": 0.4912,
+      "step": 9505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566794091147476e-05,
+      "loss": 0.4754,
+      "step": 9506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566701756291118e-05,
+      "loss": 0.5315,
+      "step": 9507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566609414316954e-05,
+      "loss": 0.5674,
+      "step": 9508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566517065226144e-05,
+      "loss": 0.4829,
+      "step": 9509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5664247090198482e-05,
+      "loss": 0.581,
+      "step": 9510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663323456992265e-05,
+      "loss": 0.5438,
+      "step": 9511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662399752654386e-05,
+      "loss": 0.5756,
+      "step": 9512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566147597719645e-05,
+      "loss": 0.4975,
+      "step": 9513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660552130630057e-05,
+      "loss": 0.5898,
+      "step": 9514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659628212966812e-05,
+      "loss": 0.5662,
+      "step": 9515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5658704224218322e-05,
+      "loss": 0.5022,
+      "step": 9516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565778016439619e-05,
+      "loss": 0.6146,
+      "step": 9517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565685603351202e-05,
+      "loss": 0.5773,
+      "step": 9518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5655931831577423e-05,
+      "loss": 0.5578,
+      "step": 9519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5655007558604006e-05,
+      "loss": 0.5662,
+      "step": 9520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654083214603373e-05,
+      "loss": 0.5875,
+      "step": 9521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565315879958714e-05,
+      "loss": 0.4698,
+      "step": 9522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565223431356692e-05,
+      "loss": 0.5698,
+      "step": 9523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5651309756554316e-05,
+      "loss": 0.4896,
+      "step": 9524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565038512856095e-05,
+      "loss": 0.5992,
+      "step": 9525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5649460429598425e-05,
+      "loss": 0.4492,
+      "step": 9526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564853565967836e-05,
+      "loss": 0.5644,
+      "step": 9527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647610818812377e-05,
+      "loss": 0.5086,
+      "step": 9528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564668590701208e-05,
+      "loss": 0.5401,
+      "step": 9529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645760924289094e-05,
+      "loss": 0.5234,
+      "step": 9530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5644835870655035e-05,
+      "loss": 0.5782,
+      "step": 9531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643910746121523e-05,
+      "loss": 0.5807,
+      "step": 9532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642985550700173e-05,
+      "loss": 0.5987,
+      "step": 9533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564206028440261e-05,
+      "loss": 0.4629,
+      "step": 9534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564113494724045e-05,
+      "loss": 0.5148,
+      "step": 9535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640209539225322e-05,
+      "loss": 0.5655,
+      "step": 9536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563928406036885e-05,
+      "loss": 0.564,
+      "step": 9537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563835851068265e-05,
+      "loss": 0.5564,
+      "step": 9538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5637432890178354e-05,
+      "loss": 0.5204,
+      "step": 9539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636507198867582e-05,
+      "loss": 0.565,
+      "step": 9540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635581436761963e-05,
+      "loss": 0.5565,
+      "step": 9541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563465560387313e-05,
+      "loss": 0.4684,
+      "step": 9542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633729700212703e-05,
+      "loss": 0.5515,
+      "step": 9543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563280372579232e-05,
+      "loss": 0.5345,
+      "step": 9544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631877680623603e-05,
+      "loss": 0.5367,
+      "step": 9545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630951564718183e-05,
+      "loss": 0.5814,
+      "step": 9546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630025378087704e-05,
+      "loss": 0.5028,
+      "step": 9547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5629099120743787e-05,
+      "loss": 0.5572,
+      "step": 9548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562817279269807e-05,
+      "loss": 0.522,
+      "step": 9549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562724639396219e-05,
+      "loss": 0.5668,
+      "step": 9550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626319924547777e-05,
+      "loss": 0.5562,
+      "step": 9551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625393384466475e-05,
+      "loss": 0.5711,
+      "step": 9552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624466773729913e-05,
+      "loss": 0.5549,
+      "step": 9553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623540092349735e-05,
+      "loss": 0.5429,
+      "step": 9554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622613340337575e-05,
+      "loss": 0.5142,
+      "step": 9555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562168651770508e-05,
+      "loss": 0.6046,
+      "step": 9556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5620759624463886e-05,
+      "loss": 0.5897,
+      "step": 9557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619832660625637e-05,
+      "loss": 0.4624,
+      "step": 9558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5618905626201975e-05,
+      "loss": 0.5525,
+      "step": 9559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617978521204545e-05,
+      "loss": 0.5098,
+      "step": 9560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617051345644987e-05,
+      "loss": 0.6228,
+      "step": 9561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561612409953495e-05,
+      "loss": 0.6024,
+      "step": 9562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5615196782886083e-05,
+      "loss": 0.5477,
+      "step": 9563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5614269395710028e-05,
+      "loss": 0.5487,
+      "step": 9564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613341938018435e-05,
+      "loss": 0.5287,
+      "step": 9565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561241440982295e-05,
+      "loss": 0.5249,
+      "step": 9566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611486811135233e-05,
+      "loss": 0.5079,
+      "step": 9567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561055914196692e-05,
+      "loss": 0.5268,
+      "step": 9568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560963140232967e-05,
+      "loss": 0.5466,
+      "step": 9569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560870359223514e-05,
+      "loss": 0.529,
+      "step": 9570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560777571169498e-05,
+      "loss": 0.5372,
+      "step": 9571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560684776072084e-05,
+      "loss": 0.5621,
+      "step": 9572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605919739324378e-05,
+      "loss": 0.573,
+      "step": 9573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604991647517253e-05,
+      "loss": 0.4995,
+      "step": 9574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604063485311112e-05,
+      "loss": 0.5322,
+      "step": 9575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5603135252717625e-05,
+      "loss": 0.505,
+      "step": 9576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602206949748446e-05,
+      "loss": 0.5559,
+      "step": 9577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560127857641523e-05,
+      "loss": 0.5059,
+      "step": 9578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5600350132729647e-05,
+      "loss": 0.6307,
+      "step": 9579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559942161870335e-05,
+      "loss": 0.5291,
+      "step": 9580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5598493034348e-05,
+      "loss": 0.4914,
+      "step": 9581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559756437967527e-05,
+      "loss": 0.529,
+      "step": 9582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559663565469681e-05,
+      "loss": 0.5968,
+      "step": 9583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.55957068594243e-05,
+      "loss": 0.572,
+      "step": 9584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5594777993869397e-05,
+      "loss": 0.5801,
+      "step": 9585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559384905804377e-05,
+      "loss": 0.5252,
+      "step": 9586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5592920051959084e-05,
+      "loss": 0.4887,
+      "step": 9587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591990975627008e-05,
+      "loss": 0.5216,
+      "step": 9588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559106182905921e-05,
+      "loss": 0.5796,
+      "step": 9589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5590132612267367e-05,
+      "loss": 0.4692,
+      "step": 9590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5589203325263143e-05,
+      "loss": 0.5598,
+      "step": 9591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588273968058215e-05,
+      "loss": 0.6442,
+      "step": 9592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558734454066425e-05,
+      "loss": 0.5001,
+      "step": 9593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586415043092925e-05,
+      "loss": 0.5637,
+      "step": 9594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585485475355914e-05,
+      "loss": 0.5688,
+      "step": 9595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5584555837464894e-05,
+      "loss": 0.548,
+      "step": 9596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558362612943154e-05,
+      "loss": 0.5468,
+      "step": 9597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558269635126753e-05,
+      "loss": 0.5854,
+      "step": 9598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581766502984536e-05,
+      "loss": 0.469,
+      "step": 9599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580836584594246e-05,
+      "loss": 0.4619,
+      "step": 9600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579906596108333e-05,
+      "loss": 0.5487,
+      "step": 9601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578976537538484e-05,
+      "loss": 0.5665,
+      "step": 9602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578046408896378e-05,
+      "loss": 0.5909,
+      "step": 9603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5577116210193693e-05,
+      "loss": 0.5431,
+      "step": 9604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557618594144212e-05,
+      "loss": 0.5417,
+      "step": 9605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575255602653337e-05,
+      "loss": 0.5934,
+      "step": 9606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557432519383903e-05,
+      "loss": 0.5993,
+      "step": 9607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557339471501089e-05,
+      "loss": 0.5778,
+      "step": 9608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55724641661806e-05,
+      "loss": 0.6061,
+      "step": 9609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5571533547359845e-05,
+      "loss": 0.5239,
+      "step": 9610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557060285856032e-05,
+      "loss": 0.5633,
+      "step": 9611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569672099793706e-05,
+      "loss": 0.5155,
+      "step": 9612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5568741271071703e-05,
+      "loss": 0.5836,
+      "step": 9613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567810372405995e-05,
+      "loss": 0.5462,
+      "step": 9614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5566879403808278e-05,
+      "loss": 0.5868,
+      "step": 9615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565948365290244e-05,
+      "loss": 0.5137,
+      "step": 9616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565017256863586e-05,
+      "loss": 0.5846,
+      "step": 9617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5564086078540002e-05,
+      "loss": 0.5415,
+      "step": 9618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563154830331183e-05,
+      "loss": 0.4971,
+      "step": 9619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562223512248827e-05,
+      "loss": 0.578,
+      "step": 9620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561292124304634e-05,
+      "loss": 0.5389,
+      "step": 9621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560360666510302e-05,
+      "loss": 0.5603,
+      "step": 9622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559429138877528e-05,
+      "loss": 0.5018,
+      "step": 9623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555849754141801e-05,
+      "loss": 0.542,
+      "step": 9624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557565874143452e-05,
+      "loss": 0.5467,
+      "step": 9625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556634137065553e-05,
+      "loss": 0.4965,
+      "step": 9626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.613,
+      "step": 9627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555477045354656e-05,
+      "loss": 0.549,
+      "step": 9628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553838507128865e-05,
+      "loss": 0.5356,
+      "step": 9629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552906490954653e-05,
+      "loss": 0.5189,
+      "step": 9630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551974405035623e-05,
+      "loss": 0.5305,
+      "step": 9631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555104224938348e-05,
+      "loss": 0.587,
+      "step": 9632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550110024009938e-05,
+      "loss": 0.501,
+      "step": 9633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5549177728926702e-05,
+      "loss": 0.4788,
+      "step": 9634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5548245364145486e-05,
+      "loss": 0.4957,
+      "step": 9635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547312929677995e-05,
+      "loss": 0.4693,
+      "step": 9636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5546380425535942e-05,
+      "loss": 0.4436,
+      "step": 9637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5545447851731043e-05,
+      "loss": 0.4951,
+      "step": 9638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544515208275005e-05,
+      "loss": 0.4559,
+      "step": 9639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543582495179546e-05,
+      "loss": 0.4726,
+      "step": 9640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554264971245638e-05,
+      "loss": 0.4729,
+      "step": 9641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554171686011723e-05,
+      "loss": 0.5122,
+      "step": 9642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55407839381738e-05,
+      "loss": 0.4573,
+      "step": 9643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5539850946637815e-05,
+      "loss": 0.4772,
+      "step": 9644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553891788552099e-05,
+      "loss": 0.4257,
+      "step": 9645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5537984754835046e-05,
+      "loss": 0.5035,
+      "step": 9646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5537051554591705e-05,
+      "loss": 0.4626,
+      "step": 9647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536118284802687e-05,
+      "loss": 0.4624,
+      "step": 9648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553518494547971e-05,
+      "loss": 0.4764,
+      "step": 9649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534251536634502e-05,
+      "loss": 0.4692,
+      "step": 9650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533318058278786e-05,
+      "loss": 0.4876,
+      "step": 9651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5532384510424284e-05,
+      "loss": 0.4623,
+      "step": 9652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5531450893082722e-05,
+      "loss": 0.4633,
+      "step": 9653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553051720626583e-05,
+      "loss": 0.5214,
+      "step": 9654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552958344998533e-05,
+      "loss": 0.4865,
+      "step": 9655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552864962425295e-05,
+      "loss": 0.4776,
+      "step": 9656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5527715729080423e-05,
+      "loss": 0.4723,
+      "step": 9657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5526781764479476e-05,
+      "loss": 0.4633,
+      "step": 9658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525847730461845e-05,
+      "loss": 0.4966,
+      "step": 9659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552491362703925e-05,
+      "loss": 0.5022,
+      "step": 9660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5523979454223436e-05,
+      "loss": 0.4618,
+      "step": 9661
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552304521202613e-05,
+      "loss": 0.5201,
+      "step": 9662
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5522110900459065e-05,
+      "loss": 0.5148,
+      "step": 9663
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521176519533978e-05,
+      "loss": 0.4709,
+      "step": 9664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520242069262605e-05,
+      "loss": 0.4636,
+      "step": 9665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519307549656683e-05,
+      "loss": 0.4918,
+      "step": 9666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518372960727947e-05,
+      "loss": 0.53,
+      "step": 9667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551743830248814e-05,
+      "loss": 0.4772,
+      "step": 9668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5516503574948998e-05,
+      "loss": 0.449,
+      "step": 9669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5515568778122263e-05,
+      "loss": 0.5144,
+      "step": 9670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514633912019676e-05,
+      "loss": 0.4951,
+      "step": 9671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513698976652974e-05,
+      "loss": 0.5006,
+      "step": 9672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551276397203391e-05,
+      "loss": 0.4672,
+      "step": 9673
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5511828898174214e-05,
+      "loss": 0.4775,
+      "step": 9674
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510893755085646e-05,
+      "loss": 0.4744,
+      "step": 9675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550995854277994e-05,
+      "loss": 0.4529,
+      "step": 9676
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5509023261268844e-05,
+      "loss": 0.5243,
+      "step": 9677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550808791056411e-05,
+      "loss": 0.4463,
+      "step": 9678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507152490677487e-05,
+      "loss": 0.4802,
+      "step": 9679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5506217001620713e-05,
+      "loss": 0.4801,
+      "step": 9680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550528144340555e-05,
+      "loss": 0.4431,
+      "step": 9681
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5504345816043742e-05,
+      "loss": 0.5001,
+      "step": 9682
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550341011954704e-05,
+      "loss": 0.4429,
+      "step": 9683
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502474353927203e-05,
+      "loss": 0.5129,
+      "step": 9684
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550153851919598e-05,
+      "loss": 0.4585,
+      "step": 9685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5500602615365124e-05,
+      "loss": 0.4725,
+      "step": 9686
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549966664244639e-05,
+      "loss": 0.4598,
+      "step": 9687
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498730600451536e-05,
+      "loss": 0.4471,
+      "step": 9688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497794489392316e-05,
+      "loss": 0.4383,
+      "step": 9689
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549685830928049e-05,
+      "loss": 0.4791,
+      "step": 9690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495922060127818e-05,
+      "loss": 0.4947,
+      "step": 9691
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494985741946054e-05,
+      "loss": 0.4962,
+      "step": 9692
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494049354746965e-05,
+      "loss": 0.433,
+      "step": 9693
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493112898542303e-05,
+      "loss": 0.5281,
+      "step": 9694
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549217637334384e-05,
+      "loss": 0.4257,
+      "step": 9695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5491239779163333e-05,
+      "loss": 0.502,
+      "step": 9696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5490303116012544e-05,
+      "loss": 0.4653,
+      "step": 9697
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489366383903245e-05,
+      "loss": 0.4447,
+      "step": 9698
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5488429582847194e-05,
+      "loss": 0.5078,
+      "step": 9699
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548749271285616e-05,
+      "loss": 0.5094,
+      "step": 9700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548655577394191e-05,
+      "loss": 0.4767,
+      "step": 9701
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485618766116214e-05,
+      "loss": 0.4319,
+      "step": 9702
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484681689390838e-05,
+      "loss": 0.5012,
+      "step": 9703
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483744543777553e-05,
+      "loss": 0.4899,
+      "step": 9704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548280732928813e-05,
+      "loss": 0.4491,
+      "step": 9705
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548187004593434e-05,
+      "loss": 0.5378,
+      "step": 9706
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480932693727953e-05,
+      "loss": 0.5319,
+      "step": 9707
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479995272680747e-05,
+      "loss": 0.4815,
+      "step": 9708
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479057782804493e-05,
+      "loss": 0.4602,
+      "step": 9709
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478120224110964e-05,
+      "loss": 0.4242,
+      "step": 9710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477182596611938e-05,
+      "loss": 0.4775,
+      "step": 9711
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5476244900319193e-05,
+      "loss": 0.4706,
+      "step": 9712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475307135244507e-05,
+      "loss": 0.4868,
+      "step": 9713
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5474369301399653e-05,
+      "loss": 0.496,
+      "step": 9714
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5473431398796418e-05,
+      "loss": 0.4562,
+      "step": 9715
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547249342744658e-05,
+      "loss": 0.5255,
+      "step": 9716
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5471555387361915e-05,
+      "loss": 0.3823,
+      "step": 9717
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547061727855421e-05,
+      "loss": 0.5244,
+      "step": 9718
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469679101035243e-05,
+      "loss": 0.5484,
+      "step": 9719
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468740854816796e-05,
+      "loss": 0.4324,
+      "step": 9720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467802539910662e-05,
+      "loss": 0.467,
+      "step": 9721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466864156328622e-05,
+      "loss": 0.4494,
+      "step": 9722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5465925704082462e-05,
+      "loss": 0.461,
+      "step": 9723
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546498718318397e-05,
+      "loss": 0.4357,
+      "step": 9724
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464048593644933e-05,
+      "loss": 0.4498,
+      "step": 9725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463109935477135e-05,
+      "loss": 0.4814,
+      "step": 9726
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462171208692377e-05,
+      "loss": 0.493,
+      "step": 9727
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546123241330244e-05,
+      "loss": 0.4662,
+      "step": 9728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460293549319117e-05,
+      "loss": 0.4394,
+      "step": 9729
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54593546167542e-05,
+      "loss": 0.5177,
+      "step": 9730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458415615619486e-05,
+      "loss": 0.4229,
+      "step": 9731
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457476545926765e-05,
+      "loss": 0.4569,
+      "step": 9732
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545653740768783e-05,
+      "loss": 0.4884,
+      "step": 9733
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455598200914487e-05,
+      "loss": 0.5167,
+      "step": 9734
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454658925618517e-05,
+      "loss": 0.4612,
+      "step": 9735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453719581811732e-05,
+      "loss": 0.491,
+      "step": 9736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545278016950592e-05,
+      "loss": 0.4443,
+      "step": 9737
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451840688712883e-05,
+      "loss": 0.4441,
+      "step": 9738
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450901139444426e-05,
+      "loss": 0.4558,
+      "step": 9739
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544996152171234e-05,
+      "loss": 0.5017,
+      "step": 9740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5449021835528435e-05,
+      "loss": 0.5081,
+      "step": 9741
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544808208090451e-05,
+      "loss": 0.4656,
+      "step": 9742
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544714225785237e-05,
+      "loss": 0.4635,
+      "step": 9743
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5446202366383817e-05,
+      "loss": 0.5313,
+      "step": 9744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445262406510656e-05,
+      "loss": 0.4645,
+      "step": 9745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5444322378244694e-05,
+      "loss": 0.4947,
+      "step": 9746
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544338228159774e-05,
+      "loss": 0.4311,
+      "step": 9747
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54424421165816e-05,
+      "loss": 0.4721,
+      "step": 9748
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5441501883208082e-05,
+      "loss": 0.4777,
+      "step": 9749
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440561581488995e-05,
+      "loss": 0.4828,
+      "step": 9750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543962121143615e-05,
+      "loss": 0.4846,
+      "step": 9751
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5438680773061357e-05,
+      "loss": 0.5113,
+      "step": 9752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543774026637643e-05,
+      "loss": 0.4389,
+      "step": 9753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543679969139318e-05,
+      "loss": 0.4817,
+      "step": 9754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543585904812342e-05,
+      "loss": 0.4111,
+      "step": 9755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543491833657897e-05,
+      "loss": 0.4563,
+      "step": 9756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433977556771638e-05,
+      "loss": 0.5067,
+      "step": 9757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433036708713244e-05,
+      "loss": 0.4859,
+      "step": 9758
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432095792415606e-05,
+      "loss": 0.5048,
+      "step": 9759
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543115480789054e-05,
+      "loss": 0.4652,
+      "step": 9760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430213755149865e-05,
+      "loss": 0.4908,
+      "step": 9761
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5429272634205404e-05,
+      "loss": 0.4938,
+      "step": 9762
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542833144506897e-05,
+      "loss": 0.5299,
+      "step": 9763
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542739018775239e-05,
+      "loss": 0.4942,
+      "step": 9764
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5426448862267485e-05,
+      "loss": 0.5217,
+      "step": 9765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425507468626083e-05,
+      "loss": 0.5112,
+      "step": 9766
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542456600684e-05,
+      "loss": 0.4937,
+      "step": 9767
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5423624476921062e-05,
+      "loss": 0.5107,
+      "step": 9768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54226828788811e-05,
+      "loss": 0.4914,
+      "step": 9769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5421741212731936e-05,
+      "loss": 0.5079,
+      "step": 9770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54207994784854e-05,
+      "loss": 0.4636,
+      "step": 9771
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419857676153317e-05,
+      "loss": 0.4535,
+      "step": 9772
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5418915805747518e-05,
+      "loss": 0.477,
+      "step": 9773
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417973867279833e-05,
+      "loss": 0.5042,
+      "step": 9774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417031860762094e-05,
+      "loss": 0.441,
+      "step": 9775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5416089786206132e-05,
+      "loss": 0.4913,
+      "step": 9776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415147643623778e-05,
+      "loss": 0.4934,
+      "step": 9777
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5414205433026867e-05,
+      "loss": 0.5168,
+      "step": 9778
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5413263154427234e-05,
+      "loss": 0.4611,
+      "step": 9779
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541232080783671e-05,
+      "loss": 0.4637,
+      "step": 9780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541137839326713e-05,
+      "loss": 0.4595,
+      "step": 9781
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5410435910730343e-05,
+      "loss": 0.435,
+      "step": 9782
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5409493360238172e-05,
+      "loss": 0.451,
+      "step": 9783
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408550741802463e-05,
+      "loss": 0.4673,
+      "step": 9784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407608055435054e-05,
+      "loss": 0.445,
+      "step": 9785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5406665301147782e-05,
+      "loss": 0.4601,
+      "step": 9786
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405722478952497e-05,
+      "loss": 0.4963,
+      "step": 9787
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5404779588861026e-05,
+      "loss": 0.5077,
+      "step": 9788
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540383663088523e-05,
+      "loss": 0.449,
+      "step": 9789
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540289360503694e-05,
+      "loss": 0.4581,
+      "step": 9790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401950511328002e-05,
+      "loss": 0.4729,
+      "step": 9791
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540100734977026e-05,
+      "loss": 0.4557,
+      "step": 9792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400064120375572e-05,
+      "loss": 0.4891,
+      "step": 9793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5399120823155774e-05,
+      "loss": 0.4566,
+      "step": 9794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539817745812271e-05,
+      "loss": 0.4615,
+      "step": 9795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397234025288242e-05,
+      "loss": 0.533,
+      "step": 9796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5396290524664205e-05,
+      "loss": 0.5015,
+      "step": 9797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539534695626246e-05,
+      "loss": 0.4736,
+      "step": 9798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539440332009486e-05,
+      "loss": 0.457,
+      "step": 9799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393459616173245e-05,
+      "loss": 0.5148,
+      "step": 9800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539251584450948e-05,
+      "loss": 0.5065,
+      "step": 9801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539157200511541e-05,
+      "loss": 0.4656,
+      "step": 9802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390628098002896e-05,
+      "loss": 0.489,
+      "step": 9803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5389684123183794e-05,
+      "loss": 0.4787,
+      "step": 9804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538874008066995e-05,
+      "loss": 0.4991,
+      "step": 9805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387795970473235e-05,
+      "loss": 0.448,
+      "step": 9806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386851792605496e-05,
+      "loss": 0.4627,
+      "step": 9807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53859075470786e-05,
+      "loss": 0.4835,
+      "step": 9808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5384963233904406e-05,
+      "loss": 0.4447,
+      "step": 9809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538401885309477e-05,
+      "loss": 0.5367,
+      "step": 9810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5383074404661555e-05,
+      "loss": 0.4861,
+      "step": 9811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382129888616623e-05,
+      "loss": 0.5048,
+      "step": 9812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538118530497184e-05,
+      "loss": 0.4736,
+      "step": 9813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380240653739068e-05,
+      "loss": 0.4964,
+      "step": 9814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5379295934930173e-05,
+      "loss": 0.4683,
+      "step": 9815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378351148557017e-05,
+      "loss": 0.4646,
+      "step": 9816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537740629463147e-05,
+      "loss": 0.4862,
+      "step": 9817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376461373165397e-05,
+      "loss": 0.4749,
+      "step": 9818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375516384170674e-05,
+      "loss": 0.4441,
+      "step": 9819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537457132765916e-05,
+      "loss": 0.4429,
+      "step": 9820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537362620364273e-05,
+      "loss": 0.4596,
+      "step": 9821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5372681012133253e-05,
+      "loss": 0.5031,
+      "step": 9822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53717357531426e-05,
+      "loss": 0.4326,
+      "step": 9823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537079042668265e-05,
+      "loss": 0.4377,
+      "step": 9824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536984503276527e-05,
+      "loss": 0.4647,
+      "step": 9825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5368899571402333e-05,
+      "loss": 0.4904,
+      "step": 9826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536795404260572e-05,
+      "loss": 0.4779,
+      "step": 9827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53670084463873e-05,
+      "loss": 0.5089,
+      "step": 9828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5366062782758953e-05,
+      "loss": 0.4823,
+      "step": 9829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365117051732564e-05,
+      "loss": 0.4696,
+      "step": 9830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364171253319997e-05,
+      "loss": 0.4538,
+      "step": 9831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363225387533142e-05,
+      "loss": 0.4565,
+      "step": 9832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362279454383875e-05,
+      "loss": 0.4116,
+      "step": 9833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361333453884074e-05,
+      "loss": 0.5214,
+      "step": 9834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536038738604563e-05,
+      "loss": 0.4782,
+      "step": 9835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535944125088042e-05,
+      "loss": 0.4969,
+      "step": 9836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358495048400325e-05,
+      "loss": 0.5259,
+      "step": 9837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357548778617233e-05,
+      "loss": 0.4936,
+      "step": 9838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356602441543026e-05,
+      "loss": 0.5628,
+      "step": 9839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355656037189597e-05,
+      "loss": 0.4494,
+      "step": 9840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5354709565568827e-05,
+      "loss": 0.4515,
+      "step": 9841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353763026692604e-05,
+      "loss": 0.4603,
+      "step": 9842
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352816420572818e-05,
+      "loss": 0.4692,
+      "step": 9843
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5351869747221356e-05,
+      "loss": 0.4466,
+      "step": 9844
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350923006650112e-05,
+      "loss": 0.4882,
+      "step": 9845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349976198870974e-05,
+      "loss": 0.487,
+      "step": 9846
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349029323895835e-05,
+      "loss": 0.4801,
+      "step": 9847
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534808238173659e-05,
+      "loss": 0.431,
+      "step": 9848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534713537240513e-05,
+      "loss": 0.4712,
+      "step": 9849
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534618829591335e-05,
+      "loss": 0.3851,
+      "step": 9850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345241152273148e-05,
+      "loss": 0.472,
+      "step": 9851
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344293941496415e-05,
+      "loss": 0.5323,
+      "step": 9852
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343346663595052e-05,
+      "loss": 0.4745,
+      "step": 9853
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5342399318580955e-05,
+      "loss": 0.5091,
+      "step": 9854
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534145190646602e-05,
+      "loss": 0.4492,
+      "step": 9855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5340504427262156e-05,
+      "loss": 0.5208,
+      "step": 9856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5339556880981253e-05,
+      "loss": 0.4831,
+      "step": 9857
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5338609267635222e-05,
+      "loss": 0.499,
+      "step": 9858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337661587235952e-05,
+      "loss": 0.438,
+      "step": 9859
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533671383979536e-05,
+      "loss": 0.4755,
+      "step": 9860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335766025325337e-05,
+      "loss": 0.4879,
+      "step": 9861
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334818143837802e-05,
+      "loss": 0.5188,
+      "step": 9862
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333870195344646e-05,
+      "loss": 0.4466,
+      "step": 9863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332922179857784e-05,
+      "loss": 0.5016,
+      "step": 9864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5331974097389118e-05,
+      "loss": 0.4529,
+      "step": 9865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533102594795056e-05,
+      "loss": 0.5128,
+      "step": 9866
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5330077731554023e-05,
+      "loss": 0.4644,
+      "step": 9867
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329129448211405e-05,
+      "loss": 0.4647,
+      "step": 9868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5328181097934625e-05,
+      "loss": 0.4953,
+      "step": 9869
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532723268073559e-05,
+      "loss": 0.4518,
+      "step": 9870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326284196626216e-05,
+      "loss": 0.5325,
+      "step": 9871
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5325335645618412e-05,
+      "loss": 0.4656,
+      "step": 9872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53243870277241e-05,
+      "loss": 0.451,
+      "step": 9873
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532343834295518e-05,
+      "loss": 0.485,
+      "step": 9874
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532248959132358e-05,
+      "loss": 0.5164,
+      "step": 9875
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5321540772841215e-05,
+      "loss": 0.4166,
+      "step": 9876
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.532059188752e-05,
+      "loss": 0.4672,
+      "step": 9877
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5319642935371853e-05,
+      "loss": 0.4842,
+      "step": 9878
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318693916408687e-05,
+      "loss": 0.4952,
+      "step": 9879
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5317744830642434e-05,
+      "loss": 0.4967,
+      "step": 9880
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5316795678085004e-05,
+      "loss": 0.467,
+      "step": 9881
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315846458748324e-05,
+      "loss": 0.5192,
+      "step": 9882
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314897172644315e-05,
+      "loss": 0.3966,
+      "step": 9883
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53139478197849e-05,
+      "loss": 0.5058,
+      "step": 9884
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5312998400182002e-05,
+      "loss": 0.4616,
+      "step": 9885
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531204891384755e-05,
+      "loss": 0.4591,
+      "step": 9886
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531109936079346e-05,
+      "loss": 0.4738,
+      "step": 9887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5310149741031668e-05,
+      "loss": 0.444,
+      "step": 9888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309200054574098e-05,
+      "loss": 0.4485,
+      "step": 9889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5308250301432677e-05,
+      "loss": 0.4878,
+      "step": 9890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307300481619334e-05,
+      "loss": 0.5348,
+      "step": 9891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306350595146002e-05,
+      "loss": 0.4915,
+      "step": 9892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530540064202461e-05,
+      "loss": 0.458,
+      "step": 9893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304450622267085e-05,
+      "loss": 0.4719,
+      "step": 9894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530350053588537e-05,
+      "loss": 0.4563,
+      "step": 9895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302550382891382e-05,
+      "loss": 0.4418,
+      "step": 9896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530160016329707e-05,
+      "loss": 0.5355,
+      "step": 9897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530064987711436e-05,
+      "loss": 0.484,
+      "step": 9898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299699524355193e-05,
+      "loss": 0.44,
+      "step": 9899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5298749105031503e-05,
+      "loss": 0.5399,
+      "step": 9900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5297798619155225e-05,
+      "loss": 0.4634,
+      "step": 9901
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296848066738302e-05,
+      "loss": 0.4149,
+      "step": 9902
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529589744779267e-05,
+      "loss": 0.4866,
+      "step": 9903
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5294946762330273e-05,
+      "loss": 0.4545,
+      "step": 9904
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5293996010363044e-05,
+      "loss": 0.4662,
+      "step": 9905
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5293045191902928e-05,
+      "loss": 0.4286,
+      "step": 9906
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529209430696187e-05,
+      "loss": 0.4646,
+      "step": 9907
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291143355551808e-05,
+      "loss": 0.4427,
+      "step": 9908
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290192337684693e-05,
+      "loss": 0.4917,
+      "step": 9909
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289241253372465e-05,
+      "loss": 0.4529,
+      "step": 9910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528829010262707e-05,
+      "loss": 0.4584,
+      "step": 9911
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5287338885460456e-05,
+      "loss": 0.4985,
+      "step": 9912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286387601884568e-05,
+      "loss": 0.4752,
+      "step": 9913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528543625191136e-05,
+      "loss": 0.4773,
+      "step": 9914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284484835552773e-05,
+      "loss": 0.4709,
+      "step": 9915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528353335282076e-05,
+      "loss": 0.5078,
+      "step": 9916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282581803727273e-05,
+      "loss": 0.4842,
+      "step": 9917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281630188284262e-05,
+      "loss": 0.4661,
+      "step": 9918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528067850650368e-05,
+      "loss": 0.464,
+      "step": 9919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279726758397485e-05,
+      "loss": 0.4623,
+      "step": 9920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5278774943977622e-05,
+      "loss": 0.5136,
+      "step": 9921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527782306325605e-05,
+      "loss": 0.503,
+      "step": 9922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527687111624472e-05,
+      "loss": 0.4637,
+      "step": 9923
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52759191029556e-05,
+      "loss": 0.4392,
+      "step": 9924
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527496702340064e-05,
+      "loss": 0.4896,
+      "step": 9925
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5274014877591797e-05,
+      "loss": 0.4824,
+      "step": 9926
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527306266554103e-05,
+      "loss": 0.4763,
+      "step": 9927
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5272110387260305e-05,
+      "loss": 0.4534,
+      "step": 9928
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5271158042761576e-05,
+      "loss": 0.4963,
+      "step": 9929
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5270205632056807e-05,
+      "loss": 0.4544,
+      "step": 9930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269253155157958e-05,
+      "loss": 0.4724,
+      "step": 9931
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268300612077e-05,
+      "loss": 0.4837,
+      "step": 9932
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267348002825885e-05,
+      "loss": 0.4682,
+      "step": 9933
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5266395327416584e-05,
+      "loss": 0.425,
+      "step": 9934
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5265442585861068e-05,
+      "loss": 0.5134,
+      "step": 9935
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5264489778171296e-05,
+      "loss": 0.4767,
+      "step": 9936
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5263536904359235e-05,
+      "loss": 0.4506,
+      "step": 9937
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526258396443686e-05,
+      "loss": 0.508,
+      "step": 9938
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526163095841613e-05,
+      "loss": 0.4136,
+      "step": 9939
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5260677886309026e-05,
+      "loss": 0.446,
+      "step": 9940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259724748127513e-05,
+      "loss": 0.4739,
+      "step": 9941
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525877154388356e-05,
+      "loss": 0.4581,
+      "step": 9942
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257818273589142e-05,
+      "loss": 0.4818,
+      "step": 9943
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525686493725623e-05,
+      "loss": 0.4124,
+      "step": 9944
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52559115348968e-05,
+      "loss": 0.4706,
+      "step": 9945
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254958066522834e-05,
+      "loss": 0.5042,
+      "step": 9946
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254004532146294e-05,
+      "loss": 0.4596,
+      "step": 9947
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253050931779166e-05,
+      "loss": 0.5217,
+      "step": 9948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252097265433421e-05,
+      "loss": 0.4509,
+      "step": 9949
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525114353312104e-05,
+      "loss": 0.4804,
+      "step": 9950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250189734854006e-05,
+      "loss": 0.4744,
+      "step": 9951
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5249235870644294e-05,
+      "loss": 0.453,
+      "step": 9952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5248281940503882e-05,
+      "loss": 0.459,
+      "step": 9953
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247327944444758e-05,
+      "loss": 0.4506,
+      "step": 9954
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246373882478899e-05,
+      "loss": 0.5213,
+      "step": 9955
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245419754618293e-05,
+      "loss": 0.541,
+      "step": 9956
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524446556087492e-05,
+      "loss": 0.5032,
+      "step": 9957
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243511301260765e-05,
+      "loss": 0.4767,
+      "step": 9958
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5242556975787812e-05,
+      "loss": 0.5115,
+      "step": 9959
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241602584468053e-05,
+      "loss": 0.4773,
+      "step": 9960
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240648127313473e-05,
+      "loss": 0.4111,
+      "step": 9961
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239693604336056e-05,
+      "loss": 0.4542,
+      "step": 9962
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238739015547797e-05,
+      "loss": 0.4494,
+      "step": 9963
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237784360960681e-05,
+      "loss": 0.4362,
+      "step": 9964
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236829640586699e-05,
+      "loss": 0.5268,
+      "step": 9965
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5235874854437845e-05,
+      "loss": 0.4497,
+      "step": 9966
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234920002526108e-05,
+      "loss": 0.4512,
+      "step": 9967
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5233965084863485e-05,
+      "loss": 0.4971,
+      "step": 9968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5233010101461964e-05,
+      "loss": 0.4889,
+      "step": 9969
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5232055052333546e-05,
+      "loss": 0.5064,
+      "step": 9970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5231099937490221e-05,
+      "loss": 0.5002,
+      "step": 9971
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230144756943992e-05,
+      "loss": 0.4845,
+      "step": 9972
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5229189510706848e-05,
+      "loss": 0.48,
+      "step": 9973
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228234198790792e-05,
+      "loss": 0.4626,
+      "step": 9974
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5227278821207823e-05,
+      "loss": 0.5141,
+      "step": 9975
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5226323377969936e-05,
+      "loss": 0.4642,
+      "step": 9976
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225367869089138e-05,
+      "loss": 0.4273,
+      "step": 9977
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5224412294577426e-05,
+      "loss": 0.4947,
+      "step": 9978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223456654446802e-05,
+      "loss": 0.5532,
+      "step": 9979
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522250094870927e-05,
+      "loss": 0.457,
+      "step": 9980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221545177376835e-05,
+      "loss": 0.4572,
+      "step": 9981
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220589340461498e-05,
+      "loss": 0.4903,
+      "step": 9982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521963343797527e-05,
+      "loss": 0.4164,
+      "step": 9983
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521867746993015e-05,
+      "loss": 0.4941,
+      "step": 9984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217721436338151e-05,
+      "loss": 0.4792,
+      "step": 9985
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216765337211276e-05,
+      "loss": 0.4902,
+      "step": 9986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521580917256154e-05,
+      "loss": 0.468,
+      "step": 9987
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5214852942400943e-05,
+      "loss": 0.5068,
+      "step": 9988
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213896646741506e-05,
+      "loss": 0.446,
+      "step": 9989
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5212940285595233e-05,
+      "loss": 0.4383,
+      "step": 9990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211983858974138e-05,
+      "loss": 0.4135,
+      "step": 9991
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211027366890237e-05,
+      "loss": 0.4876,
+      "step": 9992
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210070809355538e-05,
+      "loss": 0.5298,
+      "step": 9993
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209114186382056e-05,
+      "loss": 0.4957,
+      "step": 9994
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520815749798181e-05,
+      "loss": 0.4521,
+      "step": 9995
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5207200744166812e-05,
+      "loss": 0.4898,
+      "step": 9996
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5206243924949085e-05,
+      "loss": 0.4689,
+      "step": 9997
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205287040340641e-05,
+      "loss": 0.4533,
+      "step": 9998
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52043300903535e-05,
+      "loss": 0.4757,
+      "step": 9999
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520337307499968e-05,
+      "loss": 0.4564,
+      "step": 10000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202415994291207e-05,
+      "loss": 0.4637,
+      "step": 10001
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5201458848240091e-05,
+      "loss": 0.4828,
+      "step": 10002
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200501636858365e-05,
+      "loss": 0.4682,
+      "step": 10003
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199544360158048e-05,
+      "loss": 0.4567,
+      "step": 10004
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519858701815116e-05,
+      "loss": 0.4664,
+      "step": 10005
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5197629610849732e-05,
+      "loss": 0.4841,
+      "step": 10006
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519667213826578e-05,
+      "loss": 0.4504,
+      "step": 10007
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195714600411338e-05,
+      "loss": 0.4662,
+      "step": 10008
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519475699729843e-05,
+      "loss": 0.4267,
+      "step": 10009
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193799328939081e-05,
+      "loss": 0.4233,
+      "step": 10010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192841595345322e-05,
+      "loss": 0.4311,
+      "step": 10011
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5191883796529183e-05,
+      "loss": 0.4945,
+      "step": 10012
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190925932502692e-05,
+      "loss": 0.5418,
+      "step": 10013
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189968003277881e-05,
+      "loss": 0.4606,
+      "step": 10014
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518901000886678e-05,
+      "loss": 0.5037,
+      "step": 10015
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188051949281422e-05,
+      "loss": 0.4436,
+      "step": 10016
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187093824533845e-05,
+      "loss": 0.4765,
+      "step": 10017
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186135634636077e-05,
+      "loss": 0.4107,
+      "step": 10018
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5185177379600153e-05,
+      "loss": 0.4883,
+      "step": 10019
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184219059438113e-05,
+      "loss": 0.4625,
+      "step": 10020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183260674161988e-05,
+      "loss": 0.4818,
+      "step": 10021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518230222378382e-05,
+      "loss": 0.4869,
+      "step": 10022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518134370831565e-05,
+      "loss": 0.4877,
+      "step": 10023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180385127769508e-05,
+      "loss": 0.4687,
+      "step": 10024
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517942648215744e-05,
+      "loss": 0.4832,
+      "step": 10025
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178467771491488e-05,
+      "loss": 0.4921,
+      "step": 10026
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177508995783685e-05,
+      "loss": 0.4967,
+      "step": 10027
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176550155046084e-05,
+      "loss": 0.4684,
+      "step": 10028
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517559124929072e-05,
+      "loss": 0.3936,
+      "step": 10029
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5174632278529644e-05,
+      "loss": 0.4552,
+      "step": 10030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173673242774894e-05,
+      "loss": 0.4462,
+      "step": 10031
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5172714142038517e-05,
+      "loss": 0.4724,
+      "step": 10032
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171754976332562e-05,
+      "loss": 0.442,
+      "step": 10033
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170795745669075e-05,
+      "loss": 0.4489,
+      "step": 10034
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169836450060106e-05,
+      "loss": 0.5022,
+      "step": 10035
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.51688770895177e-05,
+      "loss": 0.4914,
+      "step": 10036
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5167917664053907e-05,
+      "loss": 0.4863,
+      "step": 10037
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516695817368078e-05,
+      "loss": 0.4641,
+      "step": 10038
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165998618410367e-05,
+      "loss": 0.4373,
+      "step": 10039
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165038998254724e-05,
+      "loss": 0.4999,
+      "step": 10040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164079313225903e-05,
+      "loss": 0.4207,
+      "step": 10041
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163119563335956e-05,
+      "loss": 0.4973,
+      "step": 10042
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5162159748596936e-05,
+      "loss": 0.4835,
+      "step": 10043
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161199869020901e-05,
+      "loss": 0.503,
+      "step": 10044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160239924619906e-05,
+      "loss": 0.5136,
+      "step": 10045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5159279915406012e-05,
+      "loss": 0.444,
+      "step": 10046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158319841391271e-05,
+      "loss": 0.4753,
+      "step": 10047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157359702587744e-05,
+      "loss": 0.4608,
+      "step": 10048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515639949900749e-05,
+      "loss": 0.4571,
+      "step": 10049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515543923066257e-05,
+      "loss": 0.4867,
+      "step": 10050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154478897565046e-05,
+      "loss": 0.5261,
+      "step": 10051
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5153518499726976e-05,
+      "loss": 0.4693,
+      "step": 10052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152558037160426e-05,
+      "loss": 0.3909,
+      "step": 10053
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151597509877457e-05,
+      "loss": 0.4744,
+      "step": 10054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5150636917890135e-05,
+      "loss": 0.4654,
+      "step": 10055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5149676261210529e-05,
+      "loss": 0.4902,
+      "step": 10056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148715539850696e-05,
+      "loss": 0.4422,
+      "step": 10057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147754753822708e-05,
+      "loss": 0.4697,
+      "step": 10058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5146793903138632e-05,
+      "loss": 0.4877,
+      "step": 10059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145832987810539e-05,
+      "loss": 0.451,
+      "step": 10060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144872007850496e-05,
+      "loss": 0.469,
+      "step": 10061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143910963270569e-05,
+      "loss": 0.4799,
+      "step": 10062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142949854082834e-05,
+      "loss": 0.5,
+      "step": 10063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141988680299357e-05,
+      "loss": 0.4866,
+      "step": 10064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141027441932217e-05,
+      "loss": 0.4845,
+      "step": 10065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140066138993485e-05,
+      "loss": 0.4665,
+      "step": 10066
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5139104771495237e-05,
+      "loss": 0.4672,
+      "step": 10067
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513814333944954e-05,
+      "loss": 0.4232,
+      "step": 10068
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513718184286848e-05,
+      "loss": 0.4906,
+      "step": 10069
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136220281764123e-05,
+      "loss": 0.4893,
+      "step": 10070
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135258656148554e-05,
+      "loss": 0.4753,
+      "step": 10071
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513429696603385e-05,
+      "loss": 0.4415,
+      "step": 10072
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513333521143209e-05,
+      "loss": 0.4557,
+      "step": 10073
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5132373392355348e-05,
+      "loss": 0.5413,
+      "step": 10074
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131411508815709e-05,
+      "loss": 0.4194,
+      "step": 10075
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130449560825255e-05,
+      "loss": 0.4698,
+      "step": 10076
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129487548396068e-05,
+      "loss": 0.542,
+      "step": 10077
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128525471540232e-05,
+      "loss": 0.5041,
+      "step": 10078
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127563330269826e-05,
+      "loss": 0.4648,
+      "step": 10079
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512660112459694e-05,
+      "loss": 0.4814,
+      "step": 10080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5125638854533651e-05,
+      "loss": 0.5113,
+      "step": 10081
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124676520092057e-05,
+      "loss": 0.5042,
+      "step": 10082
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512371412128424e-05,
+      "loss": 0.427,
+      "step": 10083
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5122751658122279e-05,
+      "loss": 0.4619,
+      "step": 10084
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121789130618276e-05,
+      "loss": 0.4542,
+      "step": 10085
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5120826538784314e-05,
+      "loss": 0.5041,
+      "step": 10086
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5119863882632481e-05,
+      "loss": 0.4604,
+      "step": 10087
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118901162174874e-05,
+      "loss": 0.4935,
+      "step": 10088
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117938377423582e-05,
+      "loss": 0.4985,
+      "step": 10089
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5116975528390695e-05,
+      "loss": 0.4569,
+      "step": 10090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511601261508831e-05,
+      "loss": 0.3998,
+      "step": 10091
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511504963752852e-05,
+      "loss": 0.4898,
+      "step": 10092
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511408659572342e-05,
+      "loss": 0.4703,
+      "step": 10093
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511312348968511e-05,
+      "loss": 0.4923,
+      "step": 10094
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112160319425675e-05,
+      "loss": 0.444,
+      "step": 10095
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5111197084957224e-05,
+      "loss": 0.4322,
+      "step": 10096
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511023378629185e-05,
+      "loss": 0.4503,
+      "step": 10097
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109270423441656e-05,
+      "loss": 0.4979,
+      "step": 10098
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108306996418738e-05,
+      "loss": 0.4712,
+      "step": 10099
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107343505235196e-05,
+      "loss": 0.5165,
+      "step": 10100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106379949903134e-05,
+      "loss": 0.4495,
+      "step": 10101
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105416330434657e-05,
+      "loss": 0.544,
+      "step": 10102
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5104452646841863e-05,
+      "loss": 0.4859,
+      "step": 10103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103488899136858e-05,
+      "loss": 0.4603,
+      "step": 10104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102525087331748e-05,
+      "loss": 0.4764,
+      "step": 10105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101561211438634e-05,
+      "loss": 0.4507,
+      "step": 10106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100597271469629e-05,
+      "loss": 0.4323,
+      "step": 10107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099633267436833e-05,
+      "loss": 0.4528,
+      "step": 10108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5098669199352362e-05,
+      "loss": 0.4839,
+      "step": 10109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097705067228314e-05,
+      "loss": 0.491,
+      "step": 10110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5096740871076811e-05,
+      "loss": 0.4785,
+      "step": 10111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095776610909951e-05,
+      "loss": 0.4723,
+      "step": 10112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094812286739858e-05,
+      "loss": 0.5072,
+      "step": 10113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093847898578635e-05,
+      "loss": 0.4709,
+      "step": 10114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5092883446438396e-05,
+      "loss": 0.4898,
+      "step": 10115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091918930331254e-05,
+      "loss": 0.4585,
+      "step": 10116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509095435026933e-05,
+      "loss": 0.5282,
+      "step": 10117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508998970626473e-05,
+      "loss": 0.5098,
+      "step": 10118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089024998329577e-05,
+      "loss": 0.4819,
+      "step": 10119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088060226475984e-05,
+      "loss": 0.4355,
+      "step": 10120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087095390716066e-05,
+      "loss": 0.4677,
+      "step": 10121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508613049106195e-05,
+      "loss": 0.4704,
+      "step": 10122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508516552752575e-05,
+      "loss": 0.4749,
+      "step": 10123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084200500119584e-05,
+      "loss": 0.5225,
+      "step": 10124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5083235408855577e-05,
+      "loss": 0.4596,
+      "step": 10125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082270253745846e-05,
+      "loss": 0.4963,
+      "step": 10126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5081305034802516e-05,
+      "loss": 0.4648,
+      "step": 10127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080339752037711e-05,
+      "loss": 0.5184,
+      "step": 10128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079374405463556e-05,
+      "loss": 0.4758,
+      "step": 10129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5078408995092173e-05,
+      "loss": 0.4713,
+      "step": 10130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5077443520935688e-05,
+      "loss": 0.4208,
+      "step": 10131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5076477983006226e-05,
+      "loss": 0.4907,
+      "step": 10132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5075512381315918e-05,
+      "loss": 0.4847,
+      "step": 10133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074546715876892e-05,
+      "loss": 0.5101,
+      "step": 10134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073580986701273e-05,
+      "loss": 0.4931,
+      "step": 10135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072615193801192e-05,
+      "loss": 0.4875,
+      "step": 10136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071649337188784e-05,
+      "loss": 0.4694,
+      "step": 10137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070683416876172e-05,
+      "loss": 0.4238,
+      "step": 10138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069717432875495e-05,
+      "loss": 0.453,
+      "step": 10139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068751385198881e-05,
+      "loss": 0.4488,
+      "step": 10140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5067785273858468e-05,
+      "loss": 0.4888,
+      "step": 10141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066819098866385e-05,
+      "loss": 0.4517,
+      "step": 10142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065852860234774e-05,
+      "loss": 0.4261,
+      "step": 10143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064886557975765e-05,
+      "loss": 0.4151,
+      "step": 10144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063920192101495e-05,
+      "loss": 0.5285,
+      "step": 10145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5062953762624111e-05,
+      "loss": 0.4899,
+      "step": 10146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061987269555735e-05,
+      "loss": 0.513,
+      "step": 10147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061020712908523e-05,
+      "loss": 0.4396,
+      "step": 10148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5060054092694603e-05,
+      "loss": 0.4832,
+      "step": 10149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505908740892612e-05,
+      "loss": 0.4651,
+      "step": 10150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5058120661615221e-05,
+      "loss": 0.4662,
+      "step": 10151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057153850774038e-05,
+      "loss": 0.4672,
+      "step": 10152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056186976414721e-05,
+      "loss": 0.4157,
+      "step": 10153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5055220038549414e-05,
+      "loss": 0.3872,
+      "step": 10154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5054253037190256e-05,
+      "loss": 0.4295,
+      "step": 10155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053285972349403e-05,
+      "loss": 0.4157,
+      "step": 10156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052318844038992e-05,
+      "loss": 0.527,
+      "step": 10157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505135165227117e-05,
+      "loss": 0.4468,
+      "step": 10158
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505038439705809e-05,
+      "loss": 0.4591,
+      "step": 10159
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049417078411903e-05,
+      "loss": 0.4789,
+      "step": 10160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048449696344753e-05,
+      "loss": 0.4994,
+      "step": 10161
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5047482250868793e-05,
+      "loss": 0.4757,
+      "step": 10162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504651474199617e-05,
+      "loss": 0.4562,
+      "step": 10163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045547169739038e-05,
+      "loss": 0.458,
+      "step": 10164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044579534109557e-05,
+      "loss": 0.4423,
+      "step": 10165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043611835119869e-05,
+      "loss": 0.4603,
+      "step": 10166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042644072782135e-05,
+      "loss": 0.4646,
+      "step": 10167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041676247108513e-05,
+      "loss": 0.5206,
+      "step": 10168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504070835811115e-05,
+      "loss": 0.5043,
+      "step": 10169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503974040580221e-05,
+      "loss": 0.4489,
+      "step": 10170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038772390193848e-05,
+      "loss": 0.4874,
+      "step": 10171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037804311298221e-05,
+      "loss": 0.4592,
+      "step": 10172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503683616912749e-05,
+      "loss": 0.4488,
+      "step": 10173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503586796369382e-05,
+      "loss": 0.4851,
+      "step": 10174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034899695009359e-05,
+      "loss": 0.4427,
+      "step": 10175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503393136308628e-05,
+      "loss": 0.5378,
+      "step": 10176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032962967936739e-05,
+      "loss": 0.5399,
+      "step": 10177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031994509572903e-05,
+      "loss": 0.52,
+      "step": 10178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031025988006935e-05,
+      "loss": 0.411,
+      "step": 10179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030057403250998e-05,
+      "loss": 0.4126,
+      "step": 10180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029088755317255e-05,
+      "loss": 0.4259,
+      "step": 10181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502812004421788e-05,
+      "loss": 0.4916,
+      "step": 10182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5027151269965035e-05,
+      "loss": 0.4979,
+      "step": 10183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026182432570886e-05,
+      "loss": 0.4691,
+      "step": 10184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025213532047605e-05,
+      "loss": 0.4949,
+      "step": 10185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502424456840736e-05,
+      "loss": 0.4306,
+      "step": 10186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023275541662324e-05,
+      "loss": 0.521,
+      "step": 10187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022306451824667e-05,
+      "loss": 0.4297,
+      "step": 10188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021337298906557e-05,
+      "loss": 0.4732,
+      "step": 10189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020368082920169e-05,
+      "loss": 0.4691,
+      "step": 10190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501939880387768e-05,
+      "loss": 0.4444,
+      "step": 10191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5018429461791256e-05,
+      "loss": 0.5141,
+      "step": 10192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501746005667308e-05,
+      "loss": 0.4537,
+      "step": 10193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5016490588535324e-05,
+      "loss": 0.455,
+      "step": 10194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015521057390166e-05,
+      "loss": 0.513,
+      "step": 10195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501455146324978e-05,
+      "loss": 0.4955,
+      "step": 10196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501358180612635e-05,
+      "loss": 0.5273,
+      "step": 10197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012612086032051e-05,
+      "loss": 0.4334,
+      "step": 10198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011642302979064e-05,
+      "loss": 0.4662,
+      "step": 10199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010672456979568e-05,
+      "loss": 0.4939,
+      "step": 10200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009702548045741e-05,
+      "loss": 0.4175,
+      "step": 10201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008732576189775e-05,
+      "loss": 0.4698,
+      "step": 10202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5007762541423848e-05,
+      "loss": 0.4856,
+      "step": 10203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500679244376014e-05,
+      "loss": 0.4843,
+      "step": 10204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005822283210841e-05,
+      "loss": 0.4641,
+      "step": 10205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004852059788129e-05,
+      "loss": 0.5255,
+      "step": 10206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.50038817735042e-05,
+      "loss": 0.4476,
+      "step": 10207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5002911424371232e-05,
+      "loss": 0.4629,
+      "step": 10208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5001941012401418e-05,
+      "loss": 0.4922,
+      "step": 10209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000970537606945e-05,
+      "loss": 0.5148,
+      "step": 10210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5169,
+      "step": 10211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4999029399592778e-05,
+      "loss": 0.4293,
+      "step": 10212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998058736397466e-05,
+      "loss": 0.4721,
+      "step": 10213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997088010426255e-05,
+      "loss": 0.4459,
+      "step": 10214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499611722169134e-05,
+      "loss": 0.4982,
+      "step": 10215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995146370204912e-05,
+      "loss": 0.4791,
+      "step": 10216
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994175455979167e-05,
+      "loss": 0.4885,
+      "step": 10217
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993204479026298e-05,
+      "loss": 0.3991,
+      "step": 10218
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992233439358499e-05,
+      "loss": 0.4769,
+      "step": 10219
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991262336987973e-05,
+      "loss": 0.4842,
+      "step": 10220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4990291171926907e-05,
+      "loss": 0.4797,
+      "step": 10221
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989319944187509e-05,
+      "loss": 0.4782,
+      "step": 10222
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498834865378197e-05,
+      "loss": 0.4855,
+      "step": 10223
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987377300722492e-05,
+      "loss": 0.475,
+      "step": 10224
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498640588502128e-05,
+      "loss": 0.5101,
+      "step": 10225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4985434406690527e-05,
+      "loss": 0.4465,
+      "step": 10226
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4984462865742441e-05,
+      "loss": 0.4931,
+      "step": 10227
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983491262189223e-05,
+      "loss": 0.4849,
+      "step": 10228
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498251959604307e-05,
+      "loss": 0.4275,
+      "step": 10229
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4981547867316199e-05,
+      "loss": 0.5026,
+      "step": 10230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980576076020805e-05,
+      "loss": 0.4868,
+      "step": 10231
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979604222169096e-05,
+      "loss": 0.4495,
+      "step": 10232
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978632305773278e-05,
+      "loss": 0.4599,
+      "step": 10233
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497766032684556e-05,
+      "loss": 0.4538,
+      "step": 10234
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497668828539815e-05,
+      "loss": 0.4962,
+      "step": 10235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975716181443258e-05,
+      "loss": 0.4808,
+      "step": 10236
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497474401499309e-05,
+      "loss": 0.4685,
+      "step": 10237
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497377178605986e-05,
+      "loss": 0.4438,
+      "step": 10238
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4972799494655776e-05,
+      "loss": 0.4327,
+      "step": 10239
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497182714079305e-05,
+      "loss": 0.4605,
+      "step": 10240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.49708547244839e-05,
+      "loss": 0.4524,
+      "step": 10241
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969882245740537e-05,
+      "loss": 0.4832,
+      "step": 10242
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968909704575171e-05,
+      "loss": 0.4875,
+      "step": 10243
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967937101000025e-05,
+      "loss": 0.4614,
+      "step": 10244
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496696443502731e-05,
+      "loss": 0.5087,
+      "step": 10245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965991706669241e-05,
+      "loss": 0.5029,
+      "step": 10246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965018915938042e-05,
+      "loss": 0.4546,
+      "step": 10247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964046062845925e-05,
+      "loss": 0.4945,
+      "step": 10248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4963073147405109e-05,
+      "loss": 0.4829,
+      "step": 10249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962100169627818e-05,
+      "loss": 0.4689,
+      "step": 10250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961127129526275e-05,
+      "loss": 0.446,
+      "step": 10251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4960154027112692e-05,
+      "loss": 0.4913,
+      "step": 10252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959180862399298e-05,
+      "loss": 0.456,
+      "step": 10253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4958207635398316e-05,
+      "loss": 0.4408,
+      "step": 10254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957234346121965e-05,
+      "loss": 0.4425,
+      "step": 10255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4956260994582478e-05,
+      "loss": 0.445,
+      "step": 10256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955287580792072e-05,
+      "loss": 0.4915,
+      "step": 10257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4954314104762976e-05,
+      "loss": 0.4872,
+      "step": 10258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495334056650742e-05,
+      "loss": 0.4543,
+      "step": 10259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4952366966037625e-05,
+      "loss": 0.4409,
+      "step": 10260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495139330336583e-05,
+      "loss": 0.4831,
+      "step": 10261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950419578504254e-05,
+      "loss": 0.4002,
+      "step": 10262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949445791465129e-05,
+      "loss": 0.4505,
+      "step": 10263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948471942260692e-05,
+      "loss": 0.407,
+      "step": 10264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947498030903167e-05,
+      "loss": 0.4933,
+      "step": 10265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494652405740479e-05,
+      "loss": 0.4454,
+      "step": 10266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4945550021777794e-05,
+      "loss": 0.4506,
+      "step": 10267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944575924034414e-05,
+      "loss": 0.4672,
+      "step": 10268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4943601764186882e-05,
+      "loss": 0.4526,
+      "step": 10269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942627542247434e-05,
+      "loss": 0.4402,
+      "step": 10270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494165325822831e-05,
+      "loss": 0.5014,
+      "step": 10271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940678912141745e-05,
+      "loss": 0.4524,
+      "step": 10272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4939704503999972e-05,
+      "loss": 0.4213,
+      "step": 10273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938730033815236e-05,
+      "loss": 0.4864,
+      "step": 10274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937755501599774e-05,
+      "loss": 0.4482,
+      "step": 10275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936780907365826e-05,
+      "loss": 0.4275,
+      "step": 10276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935806251125635e-05,
+      "loss": 0.4549,
+      "step": 10277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493483153289144e-05,
+      "loss": 0.4317,
+      "step": 10278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4933856752675483e-05,
+      "loss": 0.4959,
+      "step": 10279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493288191049001e-05,
+      "loss": 0.4603,
+      "step": 10280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931907006347266e-05,
+      "loss": 0.4357,
+      "step": 10281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930932040259492e-05,
+      "loss": 0.464,
+      "step": 10282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929957012238933e-05,
+      "loss": 0.4557,
+      "step": 10283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928981922297842e-05,
+      "loss": 0.4745,
+      "step": 10284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492800677044846e-05,
+      "loss": 0.483,
+      "step": 10285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4927031556703036e-05,
+      "loss": 0.4254,
+      "step": 10286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926056281073822e-05,
+      "loss": 0.5082,
+      "step": 10287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925080943573064e-05,
+      "loss": 0.4974,
+      "step": 10288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4924105544213014e-05,
+      "loss": 0.4724,
+      "step": 10289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923130083005922e-05,
+      "loss": 0.4812,
+      "step": 10290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922154559964042e-05,
+      "loss": 0.4474,
+      "step": 10291
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4921178975099621e-05,
+      "loss": 0.3963,
+      "step": 10292
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920203328424923e-05,
+      "loss": 0.4709,
+      "step": 10293
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919227619952188e-05,
+      "loss": 0.4656,
+      "step": 10294
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918251849693683e-05,
+      "loss": 0.4887,
+      "step": 10295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491727601766166e-05,
+      "loss": 0.5041,
+      "step": 10296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4916300123868372e-05,
+      "loss": 0.4475,
+      "step": 10297
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915324168326082e-05,
+      "loss": 0.4989,
+      "step": 10298
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914348151047044e-05,
+      "loss": 0.4365,
+      "step": 10299
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913372072043517e-05,
+      "loss": 0.5111,
+      "step": 10300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912395931327761e-05,
+      "loss": 0.4525,
+      "step": 10301
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491141972891204e-05,
+      "loss": 0.4603,
+      "step": 10302
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910443464808611e-05,
+      "loss": 0.5316,
+      "step": 10303
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909467139029736e-05,
+      "loss": 0.5095,
+      "step": 10304
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908490751587681e-05,
+      "loss": 0.5061,
+      "step": 10305
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4907514302494704e-05,
+      "loss": 0.4863,
+      "step": 10306
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906537791763073e-05,
+      "loss": 0.4999,
+      "step": 10307
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905561219405055e-05,
+      "loss": 0.4527,
+      "step": 10308
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4904584585432912e-05,
+      "loss": 0.4822,
+      "step": 10309
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4903607889858913e-05,
+      "loss": 0.4803,
+      "step": 10310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902631132695323e-05,
+      "loss": 0.4648,
+      "step": 10311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490165431395441e-05,
+      "loss": 0.4709,
+      "step": 10312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900677433648444e-05,
+      "loss": 0.4839,
+      "step": 10313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.48997004917897e-05,
+      "loss": 0.5075,
+      "step": 10314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4898723488390439e-05,
+      "loss": 0.4516,
+      "step": 10315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897746423462938e-05,
+      "loss": 0.4541,
+      "step": 10316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896769297019467e-05,
+      "loss": 0.4162,
+      "step": 10317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895792109072301e-05,
+      "loss": 0.4357,
+      "step": 10318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894814859633711e-05,
+      "loss": 0.5064,
+      "step": 10319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489383754871597e-05,
+      "loss": 0.4617,
+      "step": 10320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4892860176331358e-05,
+      "loss": 0.4519,
+      "step": 10321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891882742492147e-05,
+      "loss": 0.4658,
+      "step": 10322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890905247210615e-05,
+      "loss": 0.5486,
+      "step": 10323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488992769049904e-05,
+      "loss": 0.4446,
+      "step": 10324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888950072369701e-05,
+      "loss": 0.4991,
+      "step": 10325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488797239283487e-05,
+      "loss": 0.5194,
+      "step": 10326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886994651906836e-05,
+      "loss": 0.4864,
+      "step": 10327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886016849597875e-05,
+      "loss": 0.4331,
+      "step": 10328
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885038985920267e-05,
+      "loss": 0.4482,
+      "step": 10329
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884061060886299e-05,
+      "loss": 0.4252,
+      "step": 10330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4883083074508247e-05,
+      "loss": 0.4619,
+      "step": 10331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.48821050267984e-05,
+      "loss": 0.4616,
+      "step": 10332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488112691776904e-05,
+      "loss": 0.4941,
+      "step": 10333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488014874743245e-05,
+      "loss": 0.4731,
+      "step": 10334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879170515800923e-05,
+      "loss": 0.4703,
+      "step": 10335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878192222886741e-05,
+      "loss": 0.5141,
+      "step": 10336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877213868702187e-05,
+      "loss": 0.4565,
+      "step": 10337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876235453259555e-05,
+      "loss": 0.4764,
+      "step": 10338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875256976571135e-05,
+      "loss": 0.4425,
+      "step": 10339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4874278438649214e-05,
+      "loss": 0.4329,
+      "step": 10340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873299839506081e-05,
+      "loss": 0.5244,
+      "step": 10341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487232117915403e-05,
+      "loss": 0.473,
+      "step": 10342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871342457605351e-05,
+      "loss": 0.4713,
+      "step": 10343
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4870363674872338e-05,
+      "loss": 0.4819,
+      "step": 10344
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869384830967288e-05,
+      "loss": 0.4475,
+      "step": 10345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868405925902489e-05,
+      "loss": 0.481,
+      "step": 10346
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867426959690239e-05,
+      "loss": 0.4257,
+      "step": 10347
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866447932342834e-05,
+      "loss": 0.4828,
+      "step": 10348
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486546884387257e-05,
+      "loss": 0.4757,
+      "step": 10349
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4864489694291746e-05,
+      "loss": 0.4425,
+      "step": 10350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863510483612658e-05,
+      "loss": 0.5318,
+      "step": 10351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4862531211847606e-05,
+      "loss": 0.5112,
+      "step": 10352
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861551879008891e-05,
+      "loss": 0.4811,
+      "step": 10353
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4860572485108808e-05,
+      "loss": 0.5073,
+      "step": 10354
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4859593030159665e-05,
+      "loss": 0.4388,
+      "step": 10355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858613514173764e-05,
+      "loss": 0.4392,
+      "step": 10356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4857633937163402e-05,
+      "loss": 0.5177,
+      "step": 10357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4856654299140887e-05,
+      "loss": 0.4574,
+      "step": 10358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855674600118521e-05,
+      "loss": 0.4356,
+      "step": 10359
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485469484010861e-05,
+      "loss": 0.454,
+      "step": 10360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485371501912346e-05,
+      "loss": 0.4383,
+      "step": 10361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485273513717538e-05,
+      "loss": 0.4611,
+      "step": 10362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4851755194276672e-05,
+      "loss": 0.4303,
+      "step": 10363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850775190439648e-05,
+      "loss": 0.4689,
+      "step": 10364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849795125676616e-05,
+      "loss": 0.4133,
+      "step": 10365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848814999999883e-05,
+      "loss": 0.4515,
+      "step": 10366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847834813421767e-05,
+      "loss": 0.5132,
+      "step": 10367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846854565954572e-05,
+      "loss": 0.4978,
+      "step": 10368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484587425761061e-05,
+      "loss": 0.4719,
+      "step": 10369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844893888402199e-05,
+      "loss": 0.5024,
+      "step": 10370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843913458341646e-05,
+      "loss": 0.5041,
+      "step": 10371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4842932967441275e-05,
+      "loss": 0.5394,
+      "step": 10372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484195241571339e-05,
+      "loss": 0.4563,
+      "step": 10373
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484097180317031e-05,
+      "loss": 0.4644,
+      "step": 10374
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839991129824356e-05,
+      "loss": 0.4749,
+      "step": 10375
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839010395687844e-05,
+      "loss": 0.3978,
+      "step": 10376
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838029600773087e-05,
+      "loss": 0.4512,
+      "step": 10377
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837048745092407e-05,
+      "loss": 0.4907,
+      "step": 10378
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836067828658124e-05,
+      "loss": 0.492,
+      "step": 10379
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4835086851482558e-05,
+      "loss": 0.5098,
+      "step": 10380
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834105813578032e-05,
+      "loss": 0.4791,
+      "step": 10381
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833124714956868e-05,
+      "loss": 0.4201,
+      "step": 10382
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832143555631385e-05,
+      "loss": 0.4669,
+      "step": 10383
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4831162335613907e-05,
+      "loss": 0.3981,
+      "step": 10384
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483018105491676e-05,
+      "loss": 0.496,
+      "step": 10385
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829199713552268e-05,
+      "loss": 0.4333,
+      "step": 10386
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482821831153276e-05,
+      "loss": 0.4808,
+      "step": 10387
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827236848870556e-05,
+      "loss": 0.4546,
+      "step": 10388
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482625532557799e-05,
+      "loss": 0.5017,
+      "step": 10389
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4825273741667385e-05,
+      "loss": 0.4725,
+      "step": 10390
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482429209715107e-05,
+      "loss": 0.4674,
+      "step": 10391
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482331039204138e-05,
+      "loss": 0.4859,
+      "step": 10392
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822328626350641e-05,
+      "loss": 0.5048,
+      "step": 10393
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821346800091182e-05,
+      "loss": 0.5007,
+      "step": 10394
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820364913275338e-05,
+      "loss": 0.5164,
+      "step": 10395
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481938296591544e-05,
+      "loss": 0.4466,
+      "step": 10396
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818400958023822e-05,
+      "loss": 0.4799,
+      "step": 10397
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481741888961282e-05,
+      "loss": 0.434,
+      "step": 10398
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4816436760694765e-05,
+      "loss": 0.4864,
+      "step": 10399
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4815454571281995e-05,
+      "loss": 0.4301,
+      "step": 10400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814472321386845e-05,
+      "loss": 0.4904,
+      "step": 10401
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813490011021654e-05,
+      "loss": 0.4785,
+      "step": 10402
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481250764019876e-05,
+      "loss": 0.4665,
+      "step": 10403
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811525208930499e-05,
+      "loss": 0.5151,
+      "step": 10404
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481054271722921e-05,
+      "loss": 0.5075,
+      "step": 10405
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480956016510724e-05,
+      "loss": 0.4985,
+      "step": 10406
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808577552576922e-05,
+      "loss": 0.4999,
+      "step": 10407
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48075948796506e-05,
+      "loss": 0.4025,
+      "step": 10408
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480661214634062e-05,
+      "loss": 0.4943,
+      "step": 10409
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805629352659319e-05,
+      "loss": 0.4708,
+      "step": 10410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4804646498619044e-05,
+      "loss": 0.4886,
+      "step": 10411
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803663584232142e-05,
+      "loss": 0.4422,
+      "step": 10412
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4802680609510955e-05,
+      "loss": 0.4684,
+      "step": 10413
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801697574467832e-05,
+      "loss": 0.4938,
+      "step": 10414
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4800714479115119e-05,
+      "loss": 0.5126,
+      "step": 10415
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799731323465164e-05,
+      "loss": 0.4779,
+      "step": 10416
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479874810753031e-05,
+      "loss": 0.4568,
+      "step": 10417
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797764831322914e-05,
+      "loss": 0.4853,
+      "step": 10418
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796781494855323e-05,
+      "loss": 0.4463,
+      "step": 10419
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795798098139888e-05,
+      "loss": 0.4186,
+      "step": 10420
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479481464118896e-05,
+      "loss": 0.4492,
+      "step": 10421
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479383112401489e-05,
+      "loss": 0.4657,
+      "step": 10422
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792847546630033e-05,
+      "loss": 0.5048,
+      "step": 10423
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791863909046744e-05,
+      "loss": 0.4964,
+      "step": 10424
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790880211277376e-05,
+      "loss": 0.43,
+      "step": 10425
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789896453334284e-05,
+      "loss": 0.3906,
+      "step": 10426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4788912635229826e-05,
+      "loss": 0.4445,
+      "step": 10427
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4787928756976353e-05,
+      "loss": 0.4799,
+      "step": 10428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478694481858623e-05,
+      "loss": 0.5348,
+      "step": 10429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785960820071812e-05,
+      "loss": 0.4377,
+      "step": 10430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784976761445453e-05,
+      "loss": 0.4719,
+      "step": 10431
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783992642719523e-05,
+      "loss": 0.4699,
+      "step": 10432
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783008463906377e-05,
+      "loss": 0.46,
+      "step": 10433
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782024225018374e-05,
+      "loss": 0.4957,
+      "step": 10434
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781039926067884e-05,
+      "loss": 0.4647,
+      "step": 10435
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780055567067262e-05,
+      "loss": 0.4841,
+      "step": 10436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477907114802887e-05,
+      "loss": 0.4636,
+      "step": 10437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4778086668965082e-05,
+      "loss": 0.4741,
+      "step": 10438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777102129888257e-05,
+      "loss": 0.4471,
+      "step": 10439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776117530810761e-05,
+      "loss": 0.5037,
+      "step": 10440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775132871744962e-05,
+      "loss": 0.4378,
+      "step": 10441
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774148152703224e-05,
+      "loss": 0.4718,
+      "step": 10442
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477316337369792e-05,
+      "loss": 0.51,
+      "step": 10443
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4772178534741417e-05,
+      "loss": 0.4598,
+      "step": 10444
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771193635846084e-05,
+      "loss": 0.4954,
+      "step": 10445
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4770208677024293e-05,
+      "loss": 0.4754,
+      "step": 10446
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769223658288411e-05,
+      "loss": 0.4356,
+      "step": 10447
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768238579650813e-05,
+      "loss": 0.4111,
+      "step": 10448
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767253441123874e-05,
+      "loss": 0.4863,
+      "step": 10449
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766268242719966e-05,
+      "loss": 0.3849,
+      "step": 10450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.476528298445146e-05,
+      "loss": 0.5111,
+      "step": 10451
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764297666330732e-05,
+      "loss": 0.4769,
+      "step": 10452
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4763312288370157e-05,
+      "loss": 0.4938,
+      "step": 10453
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762326850582114e-05,
+      "loss": 0.4979,
+      "step": 10454
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476134135297898e-05,
+      "loss": 0.4525,
+      "step": 10455
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760355795573131e-05,
+      "loss": 0.4731,
+      "step": 10456
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759370178376949e-05,
+      "loss": 0.4934,
+      "step": 10457
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758384501402809e-05,
+      "loss": 0.4613,
+      "step": 10458
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4757398764663091e-05,
+      "loss": 0.4477,
+      "step": 10459
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756412968170181e-05,
+      "loss": 0.4744,
+      "step": 10460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4755427111936458e-05,
+      "loss": 0.4847,
+      "step": 10461
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754441195974302e-05,
+      "loss": 0.5001,
+      "step": 10462
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753455220296102e-05,
+      "loss": 0.5246,
+      "step": 10463
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4752469184914235e-05,
+      "loss": 0.4284,
+      "step": 10464
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475148308984109e-05,
+      "loss": 0.4583,
+      "step": 10465
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750496935089049e-05,
+      "loss": 0.4539,
+      "step": 10466
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4749510720670506e-05,
+      "loss": 0.4314,
+      "step": 10467
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748524446597835e-05,
+      "loss": 0.4454,
+      "step": 10468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747538112883437e-05,
+      "loss": 0.4794,
+      "step": 10469
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4746551719539692e-05,
+      "loss": 0.4525,
+      "step": 10470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474556526657899e-05,
+      "loss": 0.4117,
+      "step": 10471
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744578754013728e-05,
+      "loss": 0.4725,
+      "step": 10472
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743592181856285e-05,
+      "loss": 0.4351,
+      "step": 10473
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474260555011906e-05,
+      "loss": 0.4236,
+      "step": 10474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741618858814446e-05,
+      "loss": 0.4603,
+      "step": 10475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474063210795483e-05,
+      "loss": 0.458,
+      "step": 10476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739645297552613e-05,
+      "loss": 0.483,
+      "step": 10477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738658427620186e-05,
+      "loss": 0.4355,
+      "step": 10478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473767149816994e-05,
+      "loss": 0.4732,
+      "step": 10479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736684509214277e-05,
+      "loss": 0.4603,
+      "step": 10480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473569746076559e-05,
+      "loss": 0.4281,
+      "step": 10481
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4734710352836279e-05,
+      "loss": 0.4286,
+      "step": 10482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473372318543874e-05,
+      "loss": 0.467,
+      "step": 10483
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732735958585373e-05,
+      "loss": 0.4598,
+      "step": 10484
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731748672288574e-05,
+      "loss": 0.5384,
+      "step": 10485
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4730761326560752e-05,
+      "loss": 0.527,
+      "step": 10486
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729773921414299e-05,
+      "loss": 0.4901,
+      "step": 10487
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728786456861623e-05,
+      "loss": 0.4447,
+      "step": 10488
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727798932915125e-05,
+      "loss": 0.4479,
+      "step": 10489
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726811349587205e-05,
+      "loss": 0.5035,
+      "step": 10490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472582370689027e-05,
+      "loss": 0.469,
+      "step": 10491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724836004836724e-05,
+      "loss": 0.4747,
+      "step": 10492
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723848243438976e-05,
+      "loss": 0.4432,
+      "step": 10493
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722860422709429e-05,
+      "loss": 0.465,
+      "step": 10494
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472187254266049e-05,
+      "loss": 0.4477,
+      "step": 10495
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4720884603304564e-05,
+      "loss": 0.5015,
+      "step": 10496
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719896604654066e-05,
+      "loss": 0.5139,
+      "step": 10497
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4718908546721405e-05,
+      "loss": 0.5359,
+      "step": 10498
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4717920429518984e-05,
+      "loss": 0.4893,
+      "step": 10499
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716932253059222e-05,
+      "loss": 0.525,
+      "step": 10500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715944017354522e-05,
+      "loss": 0.486,
+      "step": 10501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714955722417303e-05,
+      "loss": 0.4616,
+      "step": 10502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.4991,
+      "step": 10503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4712978954894959e-05,
+      "loss": 0.4742,
+      "step": 10504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711990482334657e-05,
+      "loss": 0.3824,
+      "step": 10505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711001950591493e-05,
+      "loss": 0.4276,
+      "step": 10506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471001335967788e-05,
+      "loss": 0.4335,
+      "step": 10507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709024709606237e-05,
+      "loss": 0.4404,
+      "step": 10508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708036000388978e-05,
+      "loss": 0.4732,
+      "step": 10509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707047232038525e-05,
+      "loss": 0.4766,
+      "step": 10510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706058404567291e-05,
+      "loss": 0.469,
+      "step": 10511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4705069517987703e-05,
+      "loss": 0.4701,
+      "step": 10512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704080572312179e-05,
+      "loss": 0.4648,
+      "step": 10513
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703091567553139e-05,
+      "loss": 0.436,
+      "step": 10514
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4702102503723004e-05,
+      "loss": 0.4603,
+      "step": 10515
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701113380834198e-05,
+      "loss": 0.4607,
+      "step": 10516
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4700124198899146e-05,
+      "loss": 0.4025,
+      "step": 10517
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469913495793027e-05,
+      "loss": 0.4066,
+      "step": 10518
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698145657939998e-05,
+      "loss": 0.4893,
+      "step": 10519
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697156298940751e-05,
+      "loss": 0.4535,
+      "step": 10520
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469616688094496e-05,
+      "loss": 0.4799,
+      "step": 10521
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695177403965046e-05,
+      "loss": 0.4849,
+      "step": 10522
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4694187868013444e-05,
+      "loss": 0.4436,
+      "step": 10523
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469319827310258e-05,
+      "loss": 0.4445,
+      "step": 10524
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692208619244881e-05,
+      "loss": 0.4874,
+      "step": 10525
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469121890645278e-05,
+      "loss": 0.5155,
+      "step": 10526
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690229134738706e-05,
+      "loss": 0.5191,
+      "step": 10527
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468923930411509e-05,
+      "loss": 0.4648,
+      "step": 10528
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688249414594369e-05,
+      "loss": 0.4941,
+      "step": 10529
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687259466188974e-05,
+      "loss": 0.4563,
+      "step": 10530
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686269458911333e-05,
+      "loss": 0.4241,
+      "step": 10531
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685279392773888e-05,
+      "loss": 0.5368,
+      "step": 10532
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684289267789069e-05,
+      "loss": 0.5024,
+      "step": 10533
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683299083969316e-05,
+      "loss": 0.429,
+      "step": 10534
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4682308841327066e-05,
+      "loss": 0.4965,
+      "step": 10535
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681318539874752e-05,
+      "loss": 0.4983,
+      "step": 10536
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680328179624816e-05,
+      "loss": 0.4264,
+      "step": 10537
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4679337760589696e-05,
+      "loss": 0.4595,
+      "step": 10538
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467834728278183e-05,
+      "loss": 0.5278,
+      "step": 10539
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677356746213664e-05,
+      "loss": 0.472,
+      "step": 10540
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676366150897635e-05,
+      "loss": 0.5403,
+      "step": 10541
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4675375496846181e-05,
+      "loss": 0.4682,
+      "step": 10542
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674384784071749e-05,
+      "loss": 0.484,
+      "step": 10543
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673394012586785e-05,
+      "loss": 0.446,
+      "step": 10544
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4672403182403731e-05,
+      "loss": 0.4483,
+      "step": 10545
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467141229353503e-05,
+      "loss": 0.4189,
+      "step": 10546
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4670421345993128e-05,
+      "loss": 0.4476,
+      "step": 10547
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4669430339790473e-05,
+      "loss": 0.4862,
+      "step": 10548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668439274939512e-05,
+      "loss": 0.5087,
+      "step": 10549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4667448151452692e-05,
+      "loss": 0.4884,
+      "step": 10550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666456969342464e-05,
+      "loss": 0.4949,
+      "step": 10551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4665465728621273e-05,
+      "loss": 0.4446,
+      "step": 10552
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466447442930157e-05,
+      "loss": 0.4902,
+      "step": 10553
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4663483071395807e-05,
+      "loss": 0.4655,
+      "step": 10554
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662491654916438e-05,
+      "loss": 0.4828,
+      "step": 10555
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661500179875912e-05,
+      "loss": 0.4573,
+      "step": 10556
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660508646286682e-05,
+      "loss": 0.5005,
+      "step": 10557
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4659517054161203e-05,
+      "loss": 0.4901,
+      "step": 10558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658525403511926e-05,
+      "loss": 0.4513,
+      "step": 10559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657533694351311e-05,
+      "loss": 0.4012,
+      "step": 10560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656541926691812e-05,
+      "loss": 0.4407,
+      "step": 10561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655550100545886e-05,
+      "loss": 0.4141,
+      "step": 10562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465455821592599e-05,
+      "loss": 0.3687,
+      "step": 10563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653566272844581e-05,
+      "loss": 0.4533,
+      "step": 10564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465257427131412e-05,
+      "loss": 0.4784,
+      "step": 10565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651582211347065e-05,
+      "loss": 0.4852,
+      "step": 10566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650590092955878e-05,
+      "loss": 0.4286,
+      "step": 10567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649597916153017e-05,
+      "loss": 0.4756,
+      "step": 10568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464860568095095e-05,
+      "loss": 0.4822,
+      "step": 10569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464761338736213e-05,
+      "loss": 0.5095,
+      "step": 10570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646621035399027e-05,
+      "loss": 0.4262,
+      "step": 10571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645628625074109e-05,
+      "loss": 0.4477,
+      "step": 10572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4644636156399828e-05,
+      "loss": 0.5019,
+      "step": 10573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4643643629388659e-05,
+      "loss": 0.4826,
+      "step": 10574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642651044053066e-05,
+      "loss": 0.5288,
+      "step": 10575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641658400405516e-05,
+      "loss": 0.4573,
+      "step": 10576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4640665698458479e-05,
+      "loss": 0.4253,
+      "step": 10577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639672938224418e-05,
+      "loss": 0.4259,
+      "step": 10578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638680119715804e-05,
+      "loss": 0.5114,
+      "step": 10579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463768724294511e-05,
+      "loss": 0.5101,
+      "step": 10580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636694307924806e-05,
+      "loss": 0.4849,
+      "step": 10581
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4635701314667362e-05,
+      "loss": 0.433,
+      "step": 10582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4634708263185247e-05,
+      "loss": 0.4589,
+      "step": 10583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633715153490936e-05,
+      "loss": 0.4741,
+      "step": 10584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4632721985596904e-05,
+      "loss": 0.4341,
+      "step": 10585
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631728759515624e-05,
+      "loss": 0.4905,
+      "step": 10586
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630735475259575e-05,
+      "loss": 0.4604,
+      "step": 10587
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629742132841226e-05,
+      "loss": 0.4666,
+      "step": 10588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4628748732273057e-05,
+      "loss": 0.3964,
+      "step": 10589
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627755273567543e-05,
+      "loss": 0.4834,
+      "step": 10590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626761756737167e-05,
+      "loss": 0.4375,
+      "step": 10591
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625768181794401e-05,
+      "loss": 0.4974,
+      "step": 10592
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462477454875173e-05,
+      "loss": 0.4115,
+      "step": 10593
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462378085762163e-05,
+      "loss": 0.4698,
+      "step": 10594
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622787108416585e-05,
+      "loss": 0.469,
+      "step": 10595
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621793301149072e-05,
+      "loss": 0.5523,
+      "step": 10596
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462079943583158e-05,
+      "loss": 0.4578,
+      "step": 10597
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619805512476585e-05,
+      "loss": 0.4458,
+      "step": 10598
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618811531096578e-05,
+      "loss": 0.4433,
+      "step": 10599
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617817491704038e-05,
+      "loss": 0.4836,
+      "step": 10600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461682339431145e-05,
+      "loss": 0.4971,
+      "step": 10601
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46158292389313e-05,
+      "loss": 0.5284,
+      "step": 10602
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461483502557608e-05,
+      "loss": 0.493,
+      "step": 10603
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613840754258272e-05,
+      "loss": 0.4734,
+      "step": 10604
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612846424990368e-05,
+      "loss": 0.4551,
+      "step": 10605
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611852037784852e-05,
+      "loss": 0.4887,
+      "step": 10606
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610857592654215e-05,
+      "loss": 0.4379,
+      "step": 10607
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609863089610952e-05,
+      "loss": 0.507,
+      "step": 10608
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460886852866755e-05,
+      "loss": 0.4918,
+      "step": 10609
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607873909836499e-05,
+      "loss": 0.4502,
+      "step": 10610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4606879233130293e-05,
+      "loss": 0.4683,
+      "step": 10611
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460588449856143e-05,
+      "loss": 0.494,
+      "step": 10612
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4604889706142394e-05,
+      "loss": 0.505,
+      "step": 10613
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460389485588569e-05,
+      "loss": 0.4588,
+      "step": 10614
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602899947803807e-05,
+      "loss": 0.4534,
+      "step": 10615
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601904981909243e-05,
+      "loss": 0.4326,
+      "step": 10616
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4600909958214496e-05,
+      "loss": 0.4589,
+      "step": 10617
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459991487673206e-05,
+      "loss": 0.4858,
+      "step": 10618
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598919737474435e-05,
+      "loss": 0.4706,
+      "step": 10619
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597924540454122e-05,
+      "loss": 0.4975,
+      "step": 10620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459692928568362e-05,
+      "loss": 0.4831,
+      "step": 10621
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595933973175427e-05,
+      "loss": 0.5399,
+      "step": 10622
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594938602942049e-05,
+      "loss": 0.3962,
+      "step": 10623
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593943174995982e-05,
+      "loss": 0.4855,
+      "step": 10624
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592947689349732e-05,
+      "loss": 0.4722,
+      "step": 10625
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591952146015802e-05,
+      "loss": 0.5243,
+      "step": 10626
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590956545006693e-05,
+      "loss": 0.47,
+      "step": 10627
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589960886334917e-05,
+      "loss": 0.5075,
+      "step": 10628
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4588965170012975e-05,
+      "loss": 0.5068,
+      "step": 10629
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587969396053372e-05,
+      "loss": 0.4953,
+      "step": 10630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586973564468616e-05,
+      "loss": 0.4205,
+      "step": 10631
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585977675271218e-05,
+      "loss": 0.4582,
+      "step": 10632
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584981728473678e-05,
+      "loss": 0.4934,
+      "step": 10633
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4583985724088514e-05,
+      "loss": 0.4499,
+      "step": 10634
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582989662128232e-05,
+      "loss": 0.4766,
+      "step": 10635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581993542605343e-05,
+      "loss": 0.4512,
+      "step": 10636
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458099736553236e-05,
+      "loss": 0.4494,
+      "step": 10637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580001130921792e-05,
+      "loss": 0.4533,
+      "step": 10638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4579004838786153e-05,
+      "loss": 0.4326,
+      "step": 10639
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578008489137961e-05,
+      "loss": 0.4447,
+      "step": 10640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457701208198972e-05,
+      "loss": 0.4845,
+      "step": 10641
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576015617353953e-05,
+      "loss": 0.4852,
+      "step": 10642
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575019095243175e-05,
+      "loss": 0.5055,
+      "step": 10643
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4574022515669899e-05,
+      "loss": 0.4732,
+      "step": 10644
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4573025878646647e-05,
+      "loss": 0.4217,
+      "step": 10645
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572029184185933e-05,
+      "loss": 0.4612,
+      "step": 10646
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571032432300274e-05,
+      "loss": 0.4788,
+      "step": 10647
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570035623002193e-05,
+      "loss": 0.486,
+      "step": 10648
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4569038756304209e-05,
+      "loss": 0.4717,
+      "step": 10649
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568041832218841e-05,
+      "loss": 0.4244,
+      "step": 10650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567044850758613e-05,
+      "loss": 0.4767,
+      "step": 10651
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566047811936047e-05,
+      "loss": 0.4869,
+      "step": 10652
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4565050715763663e-05,
+      "loss": 0.4811,
+      "step": 10653
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4564053562253985e-05,
+      "loss": 0.4366,
+      "step": 10654
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456305635141954e-05,
+      "loss": 0.4577,
+      "step": 10655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562059083272853e-05,
+      "loss": 0.4354,
+      "step": 10656
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4561061757826446e-05,
+      "loss": 0.5103,
+      "step": 10657
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456006437509285e-05,
+      "loss": 0.4682,
+      "step": 10658
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455906693508459e-05,
+      "loss": 0.4784,
+      "step": 10659
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4558069437814192e-05,
+      "loss": 0.4221,
+      "step": 10660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455707188329419e-05,
+      "loss": 0.4453,
+      "step": 10661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556074271537106e-05,
+      "loss": 0.4444,
+      "step": 10662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555076602555476e-05,
+      "loss": 0.4765,
+      "step": 10663
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554078876361827e-05,
+      "loss": 0.4712,
+      "step": 10664
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553081092968696e-05,
+      "loss": 0.4452,
+      "step": 10665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552083252388607e-05,
+      "loss": 0.4552,
+      "step": 10666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551085354634102e-05,
+      "loss": 0.5046,
+      "step": 10667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550087399717707e-05,
+      "loss": 0.4751,
+      "step": 10668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454908938765196e-05,
+      "loss": 0.4818,
+      "step": 10669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548091318449395e-05,
+      "loss": 0.502,
+      "step": 10670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547093192122549e-05,
+      "loss": 0.442,
+      "step": 10671
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454609500868396e-05,
+      "loss": 0.4232,
+      "step": 10672
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4545096768146158e-05,
+      "loss": 0.4972,
+      "step": 10673
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544098470521691e-05,
+      "loss": 0.4162,
+      "step": 10674
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454310011582309e-05,
+      "loss": 0.4768,
+      "step": 10675
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542101704062897e-05,
+      "loss": 0.4902,
+      "step": 10676
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541103235253655e-05,
+      "loss": 0.4795,
+      "step": 10677
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4540104709407899e-05,
+      "loss": 0.4362,
+      "step": 10678
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539106126538173e-05,
+      "loss": 0.4802,
+      "step": 10679
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538107486657023e-05,
+      "loss": 0.4901,
+      "step": 10680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4537108789776983e-05,
+      "loss": 0.4628,
+      "step": 10681
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536110035910611e-05,
+      "loss": 0.4357,
+      "step": 10682
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453511122507044e-05,
+      "loss": 0.4306,
+      "step": 10683
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4534112357269012e-05,
+      "loss": 0.5227,
+      "step": 10684
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533113432518886e-05,
+      "loss": 0.4357,
+      "step": 10685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532114450832598e-05,
+      "loss": 0.4706,
+      "step": 10686
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45311154122227e-05,
+      "loss": 0.4878,
+      "step": 10687
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4530116316701738e-05,
+      "loss": 0.4397,
+      "step": 10688
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4529117164282263e-05,
+      "loss": 0.4507,
+      "step": 10689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528117954976819e-05,
+      "loss": 0.4847,
+      "step": 10690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527118688797963e-05,
+      "loss": 0.4575,
+      "step": 10691
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4526119365758242e-05,
+      "loss": 0.4429,
+      "step": 10692
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525119985870209e-05,
+      "loss": 0.4737,
+      "step": 10693
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524120549146415e-05,
+      "loss": 0.4671,
+      "step": 10694
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452312105559941e-05,
+      "loss": 0.5017,
+      "step": 10695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522121505241755e-05,
+      "loss": 0.4591,
+      "step": 10696
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521121898086001e-05,
+      "loss": 0.4557,
+      "step": 10697
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520122234144703e-05,
+      "loss": 0.5046,
+      "step": 10698
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519122513430415e-05,
+      "loss": 0.4572,
+      "step": 10699
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4518122735955695e-05,
+      "loss": 0.4598,
+      "step": 10700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45171229017331e-05,
+      "loss": 0.4746,
+      "step": 10701
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451612301077519e-05,
+      "loss": 0.4821,
+      "step": 10702
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515123063094521e-05,
+      "loss": 0.4532,
+      "step": 10703
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514123058703656e-05,
+      "loss": 0.4612,
+      "step": 10704
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451312299761515e-05,
+      "loss": 0.4137,
+      "step": 10705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512122879841565e-05,
+      "loss": 0.4488,
+      "step": 10706
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511122705395467e-05,
+      "loss": 0.4927,
+      "step": 10707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510122474289413e-05,
+      "loss": 0.4763,
+      "step": 10708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450912218653597e-05,
+      "loss": 0.4552,
+      "step": 10709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508121842147698e-05,
+      "loss": 0.4753,
+      "step": 10710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507121441137164e-05,
+      "loss": 0.4464,
+      "step": 10711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4506120983516931e-05,
+      "loss": 0.526,
+      "step": 10712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505120469299567e-05,
+      "loss": 0.4674,
+      "step": 10713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450411989849764e-05,
+      "loss": 0.5245,
+      "step": 10714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503119271123712e-05,
+      "loss": 0.4682,
+      "step": 10715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4502118587190355e-05,
+      "loss": 0.4535,
+      "step": 10716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501117846710135e-05,
+      "loss": 0.4806,
+      "step": 10717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500117049695623e-05,
+      "loss": 0.5277,
+      "step": 10718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449911619615939e-05,
+      "loss": 0.4613,
+      "step": 10719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498115286114005e-05,
+      "loss": 0.4875,
+      "step": 10720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497114319572042e-05,
+      "loss": 0.4915,
+      "step": 10721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496113296546068e-05,
+      "loss": 0.4553,
+      "step": 10722
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449511221704866e-05,
+      "loss": 0.4727,
+      "step": 10723
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4494111081092392e-05,
+      "loss": 0.5085,
+      "step": 10724
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493109888689838e-05,
+      "loss": 0.4422,
+      "step": 10725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449210863985357e-05,
+      "loss": 0.4576,
+      "step": 10726
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4491107334596164e-05,
+      "loss": 0.504,
+      "step": 10727
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490105972930201e-05,
+      "loss": 0.551,
+      "step": 10728
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4489104554868255e-05,
+      "loss": 0.4821,
+      "step": 10729
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488103080422906e-05,
+      "loss": 0.4915,
+      "step": 10730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448710154960673e-05,
+      "loss": 0.504,
+      "step": 10731
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486099962432305e-05,
+      "loss": 0.4975,
+      "step": 10732
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485098318912217e-05,
+      "loss": 0.4989,
+      "step": 10733
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448409661905904e-05,
+      "loss": 0.4828,
+      "step": 10734
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4483094862885361e-05,
+      "loss": 0.4205,
+      "step": 10735
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482093050403758e-05,
+      "loss": 0.4893,
+      "step": 10736
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4481091181626816e-05,
+      "loss": 0.4819,
+      "step": 10737
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4480089256567114e-05,
+      "loss": 0.4666,
+      "step": 10738
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479087275237245e-05,
+      "loss": 0.4657,
+      "step": 10739
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4478085237649786e-05,
+      "loss": 0.4723,
+      "step": 10740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477083143817327e-05,
+      "loss": 0.4636,
+      "step": 10741
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4476080993752453e-05,
+      "loss": 0.4544,
+      "step": 10742
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4475078787467748e-05,
+      "loss": 0.4118,
+      "step": 10743
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474076524975807e-05,
+      "loss": 0.4626,
+      "step": 10744
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473074206289212e-05,
+      "loss": 0.4953,
+      "step": 10745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472071831420559e-05,
+      "loss": 0.4456,
+      "step": 10746
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471069400382426e-05,
+      "loss": 0.4972,
+      "step": 10747
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4470066913187416e-05,
+      "loss": 0.4683,
+      "step": 10748
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469064369848114e-05,
+      "loss": 0.4643,
+      "step": 10749
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4468061770377109e-05,
+      "loss": 0.4585,
+      "step": 10750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467059114787003e-05,
+      "loss": 0.5009,
+      "step": 10751
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466056403090382e-05,
+      "loss": 0.5158,
+      "step": 10752
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465053635299843e-05,
+      "loss": 0.4568,
+      "step": 10753
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4464050811427983e-05,
+      "loss": 0.5116,
+      "step": 10754
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4463047931487391e-05,
+      "loss": 0.4722,
+      "step": 10755
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462044995490669e-05,
+      "loss": 0.4667,
+      "step": 10756
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461042003450412e-05,
+      "loss": 0.4596,
+      "step": 10757
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4460038955379216e-05,
+      "loss": 0.4699,
+      "step": 10758
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4459035851289682e-05,
+      "loss": 0.4907,
+      "step": 10759
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458032691194408e-05,
+      "loss": 0.486,
+      "step": 10760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457029475105994e-05,
+      "loss": 0.4625,
+      "step": 10761
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456026203037041e-05,
+      "loss": 0.5148,
+      "step": 10762
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455022875000149e-05,
+      "loss": 0.4737,
+      "step": 10763
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454019491007918e-05,
+      "loss": 0.4663,
+      "step": 10764
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453016051072955e-05,
+      "loss": 0.4723,
+      "step": 10765
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445201255520786e-05,
+      "loss": 0.3934,
+      "step": 10766
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451009003425238e-05,
+      "loss": 0.4508,
+      "step": 10767
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450005395737694e-05,
+      "loss": 0.4869,
+      "step": 10768
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444900173215783e-05,
+      "loss": 0.4565,
+      "step": 10769
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444799801269826e-05,
+      "loss": 0.4409,
+      "step": 10770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446994237371582e-05,
+      "loss": 0.4462,
+      "step": 10771
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444599040619041e-05,
+      "loss": 0.3791,
+      "step": 10772
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4444986519167347e-05,
+      "loss": 0.4941,
+      "step": 10773
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443982576315007e-05,
+      "loss": 0.4477,
+      "step": 10774
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442978577645996e-05,
+      "loss": 0.4751,
+      "step": 10775
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441974523172924e-05,
+      "loss": 0.4636,
+      "step": 10776
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4440970412908407e-05,
+      "loss": 0.4625,
+      "step": 10777
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439966246865046e-05,
+      "loss": 0.4809,
+      "step": 10778
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4438962025055468e-05,
+      "loss": 0.505,
+      "step": 10779
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437957747492273e-05,
+      "loss": 0.4831,
+      "step": 10780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4436953414188082e-05,
+      "loss": 0.5189,
+      "step": 10781
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435949025155509e-05,
+      "loss": 0.5293,
+      "step": 10782
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4434944580407167e-05,
+      "loss": 0.494,
+      "step": 10783
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443394007995567e-05,
+      "loss": 0.4848,
+      "step": 10784
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432935523813641e-05,
+      "loss": 0.4769,
+      "step": 10785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431930911993688e-05,
+      "loss": 0.4781,
+      "step": 10786
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4430926244508441e-05,
+      "loss": 0.4492,
+      "step": 10787
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429921521370508e-05,
+      "loss": 0.4584,
+      "step": 10788
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4428916742592516e-05,
+      "loss": 0.4725,
+      "step": 10789
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427911908187077e-05,
+      "loss": 0.5298,
+      "step": 10790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442690701816682e-05,
+      "loss": 0.4801,
+      "step": 10791
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425902072544364e-05,
+      "loss": 0.436,
+      "step": 10792
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4424897071332329e-05,
+      "loss": 0.4813,
+      "step": 10793
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423892014543336e-05,
+      "loss": 0.4363,
+      "step": 10794
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.3999,
+      "step": 10795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421881734284986e-05,
+      "loss": 0.4713,
+      "step": 10796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420876510840873e-05,
+      "loss": 0.4842,
+      "step": 10797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419871231870305e-05,
+      "loss": 0.422,
+      "step": 10798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4418865897385908e-05,
+      "loss": 0.4904,
+      "step": 10799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4417860507400305e-05,
+      "loss": 0.4946,
+      "step": 10800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4416855061926128e-05,
+      "loss": 0.469,
+      "step": 10801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415849560976003e-05,
+      "loss": 0.5176,
+      "step": 10802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414844004562563e-05,
+      "loss": 0.4465,
+      "step": 10803
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413838392698431e-05,
+      "loss": 0.4801,
+      "step": 10804
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412832725396246e-05,
+      "loss": 0.4603,
+      "step": 10805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441182700266863e-05,
+      "loss": 0.5363,
+      "step": 10806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410821224528224e-05,
+      "loss": 0.4724,
+      "step": 10807
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4409815390987654e-05,
+      "loss": 0.4601,
+      "step": 10808
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408809502059558e-05,
+      "loss": 0.4674,
+      "step": 10809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407803557756565e-05,
+      "loss": 0.4505,
+      "step": 10810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4406797558091314e-05,
+      "loss": 0.507,
+      "step": 10811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440579150307644e-05,
+      "loss": 0.4655,
+      "step": 10812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4404785392724573e-05,
+      "loss": 0.4398,
+      "step": 10813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403779227048362e-05,
+      "loss": 0.5112,
+      "step": 10814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402773006060433e-05,
+      "loss": 0.4924,
+      "step": 10815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401766729773428e-05,
+      "loss": 0.4579,
+      "step": 10816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400760398199989e-05,
+      "loss": 0.5119,
+      "step": 10817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4399754011352748e-05,
+      "loss": 0.4996,
+      "step": 10818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398747569244355e-05,
+      "loss": 0.493,
+      "step": 10819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4397741071887446e-05,
+      "loss": 0.4824,
+      "step": 10820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439673451929466e-05,
+      "loss": 0.4765,
+      "step": 10821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395727911478643e-05,
+      "loss": 0.4623,
+      "step": 10822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4394721248452037e-05,
+      "loss": 0.4535,
+      "step": 10823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439371453022749e-05,
+      "loss": 0.4471,
+      "step": 10824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439270775681764e-05,
+      "loss": 0.4881,
+      "step": 10825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4391700928235133e-05,
+      "loss": 0.4525,
+      "step": 10826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390694044492618e-05,
+      "loss": 0.3961,
+      "step": 10827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389687105602738e-05,
+      "loss": 0.4592,
+      "step": 10828
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388680111578144e-05,
+      "loss": 0.4531,
+      "step": 10829
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4387673062431482e-05,
+      "loss": 0.4503,
+      "step": 10830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4386665958175398e-05,
+      "loss": 0.5156,
+      "step": 10831
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4385658798822546e-05,
+      "loss": 0.4698,
+      "step": 10832
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384651584385573e-05,
+      "loss": 0.3905,
+      "step": 10833
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383644314877132e-05,
+      "loss": 0.4599,
+      "step": 10834
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382636990309873e-05,
+      "loss": 0.4535,
+      "step": 10835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381629610696446e-05,
+      "loss": 0.4686,
+      "step": 10836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4380622176049507e-05,
+      "loss": 0.5166,
+      "step": 10837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4379614686381706e-05,
+      "loss": 0.4642,
+      "step": 10838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4378607141705702e-05,
+      "loss": 0.4306,
+      "step": 10839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377599542034147e-05,
+      "loss": 0.4917,
+      "step": 10840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376591887379696e-05,
+      "loss": 0.4417,
+      "step": 10841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375584177755007e-05,
+      "loss": 0.4594,
+      "step": 10842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4374576413172731e-05,
+      "loss": 0.5042,
+      "step": 10843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373568593645533e-05,
+      "loss": 0.4631,
+      "step": 10844
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372560719186071e-05,
+      "loss": 0.4927,
+      "step": 10845
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4371552789807e-05,
+      "loss": 0.527,
+      "step": 10846
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437054480552098e-05,
+      "loss": 0.5,
+      "step": 10847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4369536766340672e-05,
+      "loss": 0.5231,
+      "step": 10848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368528672278739e-05,
+      "loss": 0.4641,
+      "step": 10849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436752052334784e-05,
+      "loss": 0.4754,
+      "step": 10850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366512319560642e-05,
+      "loss": 0.4688,
+      "step": 10851
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43655040609298e-05,
+      "loss": 0.4449,
+      "step": 10852
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364495747467984e-05,
+      "loss": 0.4564,
+      "step": 10853
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436348737918786e-05,
+      "loss": 0.4361,
+      "step": 10854
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362478956102087e-05,
+      "loss": 0.4546,
+      "step": 10855
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361470478223334e-05,
+      "loss": 0.4793,
+      "step": 10856
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360461945564267e-05,
+      "loss": 0.4499,
+      "step": 10857
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4359453358137556e-05,
+      "loss": 0.5017,
+      "step": 10858
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358444715955866e-05,
+      "loss": 0.5485,
+      "step": 10859
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357436019031862e-05,
+      "loss": 0.4218,
+      "step": 10860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356427267378226e-05,
+      "loss": 0.5061,
+      "step": 10861
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355418461007615e-05,
+      "loss": 0.4527,
+      "step": 10862
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354409599932702e-05,
+      "loss": 0.4374,
+      "step": 10863
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353400684166162e-05,
+      "loss": 0.3649,
+      "step": 10864
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4352391713720668e-05,
+      "loss": 0.4806,
+      "step": 10865
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351382688608887e-05,
+      "loss": 0.5022,
+      "step": 10866
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43503736088435e-05,
+      "loss": 0.4675,
+      "step": 10867
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349364474437172e-05,
+      "loss": 0.4947,
+      "step": 10868
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348355285402585e-05,
+      "loss": 0.4634,
+      "step": 10869
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347346041752412e-05,
+      "loss": 0.4247,
+      "step": 10870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434633674349933e-05,
+      "loss": 0.4988,
+      "step": 10871
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345327390656014e-05,
+      "loss": 0.4509,
+      "step": 10872
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4344317983235143e-05,
+      "loss": 0.4295,
+      "step": 10873
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343308521249396e-05,
+      "loss": 0.477,
+      "step": 10874
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342299004711447e-05,
+      "loss": 0.47,
+      "step": 10875
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434128943363398e-05,
+      "loss": 0.4654,
+      "step": 10876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340279808029679e-05,
+      "loss": 0.4841,
+      "step": 10877
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339270127911218e-05,
+      "loss": 0.5223,
+      "step": 10878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338260393291279e-05,
+      "loss": 0.4592,
+      "step": 10879
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337250604182549e-05,
+      "loss": 0.4365,
+      "step": 10880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336240760597707e-05,
+      "loss": 0.4372,
+      "step": 10881
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335230862549439e-05,
+      "loss": 0.4616,
+      "step": 10882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4334220910050429e-05,
+      "loss": 0.4851,
+      "step": 10883
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333210903113361e-05,
+      "loss": 0.496,
+      "step": 10884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332200841750922e-05,
+      "loss": 0.5502,
+      "step": 10885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331190725975797e-05,
+      "loss": 0.4635,
+      "step": 10886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330180555800674e-05,
+      "loss": 0.5008,
+      "step": 10887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329170331238244e-05,
+      "loss": 0.4042,
+      "step": 10888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328160052301188e-05,
+      "loss": 0.4925,
+      "step": 10889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43271497190022e-05,
+      "loss": 0.4785,
+      "step": 10890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432613933135397e-05,
+      "loss": 0.4583,
+      "step": 10891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325128889369186e-05,
+      "loss": 0.4414,
+      "step": 10892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324118393060545e-05,
+      "loss": 0.5055,
+      "step": 10893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432310784244073e-05,
+      "loss": 0.4781,
+      "step": 10894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322097237522441e-05,
+      "loss": 0.4761,
+      "step": 10895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4321086578318368e-05,
+      "loss": 0.4784,
+      "step": 10896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320075864841205e-05,
+      "loss": 0.4577,
+      "step": 10897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4319065097103648e-05,
+      "loss": 0.4536,
+      "step": 10898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318054275118394e-05,
+      "loss": 0.4547,
+      "step": 10899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317043398898131e-05,
+      "loss": 0.4655,
+      "step": 10900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316032468455565e-05,
+      "loss": 0.4397,
+      "step": 10901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4315021483803388e-05,
+      "loss": 0.454,
+      "step": 10902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314010444954301e-05,
+      "loss": 0.4546,
+      "step": 10903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312999351920999e-05,
+      "loss": 0.4562,
+      "step": 10904
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311988204716184e-05,
+      "loss": 0.4612,
+      "step": 10905
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310977003352558e-05,
+      "loss": 0.4167,
+      "step": 10906
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309965747842817e-05,
+      "loss": 0.4743,
+      "step": 10907
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4308954438199667e-05,
+      "loss": 0.4789,
+      "step": 10908
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307943074435807e-05,
+      "loss": 0.4131,
+      "step": 10909
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4306931656563938e-05,
+      "loss": 0.4606,
+      "step": 10910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305920184596771e-05,
+      "loss": 0.5018,
+      "step": 10911
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304908658547001e-05,
+      "loss": 0.4647,
+      "step": 10912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303897078427338e-05,
+      "loss": 0.4734,
+      "step": 10913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302885444250487e-05,
+      "loss": 0.5177,
+      "step": 10914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301873756029156e-05,
+      "loss": 0.4685,
+      "step": 10915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300862013776049e-05,
+      "loss": 0.4825,
+      "step": 10916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4299850217503873e-05,
+      "loss": 0.4932,
+      "step": 10917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4298838367225335e-05,
+      "loss": 0.4743,
+      "step": 10918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297826462953152e-05,
+      "loss": 0.4978,
+      "step": 10919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4296814504700023e-05,
+      "loss": 0.4766,
+      "step": 10920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295802492478666e-05,
+      "loss": 0.4421,
+      "step": 10921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429479042630179e-05,
+      "loss": 0.5238,
+      "step": 10922
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293778306182102e-05,
+      "loss": 0.4788,
+      "step": 10923
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429276613213232e-05,
+      "loss": 0.5129,
+      "step": 10924
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4291753904165158e-05,
+      "loss": 0.4609,
+      "step": 10925
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429074162229332e-05,
+      "loss": 0.4604,
+      "step": 10926
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289729286529532e-05,
+      "loss": 0.4349,
+      "step": 10927
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288716896886503e-05,
+      "loss": 0.4686,
+      "step": 10928
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428770445337695e-05,
+      "loss": 0.4368,
+      "step": 10929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286691956013587e-05,
+      "loss": 0.4595,
+      "step": 10930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285679404809135e-05,
+      "loss": 0.4642,
+      "step": 10931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284666799776306e-05,
+      "loss": 0.481,
+      "step": 10932
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4283654140927825e-05,
+      "loss": 0.5062,
+      "step": 10933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282641428276409e-05,
+      "loss": 0.4795,
+      "step": 10934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4281628661834774e-05,
+      "loss": 0.4612,
+      "step": 10935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280615841615646e-05,
+      "loss": 0.4882,
+      "step": 10936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427960296763174e-05,
+      "loss": 0.491,
+      "step": 10937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4278590039895784e-05,
+      "loss": 0.4422,
+      "step": 10938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4277577058420495e-05,
+      "loss": 0.5031,
+      "step": 10939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276564023218603e-05,
+      "loss": 0.4652,
+      "step": 10940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275550934302822e-05,
+      "loss": 0.4615,
+      "step": 10941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274537791685885e-05,
+      "loss": 0.4871,
+      "step": 10942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273524595380513e-05,
+      "loss": 0.4592,
+      "step": 10943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272511345399432e-05,
+      "loss": 0.4677,
+      "step": 10944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427149804175537e-05,
+      "loss": 0.4794,
+      "step": 10945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270484684461054e-05,
+      "loss": 0.4944,
+      "step": 10946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269471273529212e-05,
+      "loss": 0.497,
+      "step": 10947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268457808972568e-05,
+      "loss": 0.46,
+      "step": 10948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4267444290803858e-05,
+      "loss": 0.5234,
+      "step": 10949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266430719035808e-05,
+      "loss": 0.5047,
+      "step": 10950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265417093681148e-05,
+      "loss": 0.4587,
+      "step": 10951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264403414752612e-05,
+      "loss": 0.4737,
+      "step": 10952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263389682262931e-05,
+      "loss": 0.4887,
+      "step": 10953
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4262375896224834e-05,
+      "loss": 0.5129,
+      "step": 10954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261362056651057e-05,
+      "loss": 0.4335,
+      "step": 10955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426034816355434e-05,
+      "loss": 0.4224,
+      "step": 10956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259334216947405e-05,
+      "loss": 0.4449,
+      "step": 10957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258320216842996e-05,
+      "loss": 0.47,
+      "step": 10958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257306163253845e-05,
+      "loss": 0.4753,
+      "step": 10959
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4256292056192692e-05,
+      "loss": 0.4584,
+      "step": 10960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4255277895672272e-05,
+      "loss": 0.4539,
+      "step": 10961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254263681705325e-05,
+      "loss": 0.5095,
+      "step": 10962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253249414304584e-05,
+      "loss": 0.5346,
+      "step": 10963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252235093482794e-05,
+      "loss": 0.5004,
+      "step": 10964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251220719252694e-05,
+      "loss": 0.4725,
+      "step": 10965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250206291627023e-05,
+      "loss": 0.4602,
+      "step": 10966
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4249191810618524e-05,
+      "loss": 0.4676,
+      "step": 10967
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248177276239937e-05,
+      "loss": 0.4706,
+      "step": 10968
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247162688504003e-05,
+      "loss": 0.4927,
+      "step": 10969
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424614804742347e-05,
+      "loss": 0.4819,
+      "step": 10970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424513335301108e-05,
+      "loss": 0.4967,
+      "step": 10971
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244118605279577e-05,
+      "loss": 0.3758,
+      "step": 10972
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243103804241708e-05,
+      "loss": 0.4419,
+      "step": 10973
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4242088949910214e-05,
+      "loss": 0.4816,
+      "step": 10974
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241074042297846e-05,
+      "loss": 0.4523,
+      "step": 10975
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240059081417355e-05,
+      "loss": 0.4335,
+      "step": 10976
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423904406728148e-05,
+      "loss": 0.4974,
+      "step": 10977
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238028999902977e-05,
+      "loss": 0.4102,
+      "step": 10978
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423701387929459e-05,
+      "loss": 0.5,
+      "step": 10979
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423599870546907e-05,
+      "loss": 0.4968,
+      "step": 10980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4234983478439173e-05,
+      "loss": 0.489,
+      "step": 10981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4233968198217645e-05,
+      "loss": 0.4874,
+      "step": 10982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232952864817237e-05,
+      "loss": 0.4245,
+      "step": 10983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4231937478250707e-05,
+      "loss": 0.4933,
+      "step": 10984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4230922038530805e-05,
+      "loss": 0.4107,
+      "step": 10985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229906545670282e-05,
+      "loss": 0.4316,
+      "step": 10986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228890999681899e-05,
+      "loss": 0.4619,
+      "step": 10987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4227875400578408e-05,
+      "loss": 0.436,
+      "step": 10988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226859748372566e-05,
+      "loss": 0.4856,
+      "step": 10989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225844043077128e-05,
+      "loss": 0.5078,
+      "step": 10990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224828284704848e-05,
+      "loss": 0.4725,
+      "step": 10991
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223812473268492e-05,
+      "loss": 0.4725,
+      "step": 10992
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222796608780817e-05,
+      "loss": 0.4192,
+      "step": 10993
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221780691254576e-05,
+      "loss": 0.4519,
+      "step": 10994
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220764720702536e-05,
+      "loss": 0.4755,
+      "step": 10995
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219748697137452e-05,
+      "loss": 0.4486,
+      "step": 10996
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421873262057209e-05,
+      "loss": 0.4972,
+      "step": 10997
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421771649101921e-05,
+      "loss": 0.4148,
+      "step": 10998
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4216700308491575e-05,
+      "loss": 0.4777,
+      "step": 10999
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215684073001946e-05,
+      "loss": 0.433,
+      "step": 11000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421466778456309e-05,
+      "loss": 0.5068,
+      "step": 11001
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421365144318777e-05,
+      "loss": 0.398,
+      "step": 11002
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4212635048888754e-05,
+      "loss": 0.4511,
+      "step": 11003
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211618601678803e-05,
+      "loss": 0.4575,
+      "step": 11004
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210602101570687e-05,
+      "loss": 0.4528,
+      "step": 11005
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4209585548577174e-05,
+      "loss": 0.4756,
+      "step": 11006
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4208568942711032e-05,
+      "loss": 0.4546,
+      "step": 11007
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4207552283985027e-05,
+      "loss": 0.4545,
+      "step": 11008
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206535572411928e-05,
+      "loss": 0.4599,
+      "step": 11009
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205518808004507e-05,
+      "loss": 0.4847,
+      "step": 11010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204501990775535e-05,
+      "loss": 0.4429,
+      "step": 11011
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203485120737783e-05,
+      "loss": 0.4542,
+      "step": 11012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202468197904023e-05,
+      "loss": 0.4299,
+      "step": 11013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201451222287026e-05,
+      "loss": 0.4598,
+      "step": 11014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200434193899569e-05,
+      "loss": 0.4844,
+      "step": 11015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199417112754419e-05,
+      "loss": 0.5186,
+      "step": 11016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198399978864354e-05,
+      "loss": 0.4721,
+      "step": 11017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197382792242153e-05,
+      "loss": 0.5054,
+      "step": 11018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196365552900591e-05,
+      "loss": 0.505,
+      "step": 11019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419534826085244e-05,
+      "loss": 0.504,
+      "step": 11020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419433091611048e-05,
+      "loss": 0.4568,
+      "step": 11021
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419331351868749e-05,
+      "loss": 0.419,
+      "step": 11022
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4192296068596245e-05,
+      "loss": 0.4776,
+      "step": 11023
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419127856584953e-05,
+      "loss": 0.4932,
+      "step": 11024
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4190261010460119e-05,
+      "loss": 0.4703,
+      "step": 11025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189243402440795e-05,
+      "loss": 0.49,
+      "step": 11026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418822574180434e-05,
+      "loss": 0.4334,
+      "step": 11027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4187208028563535e-05,
+      "loss": 0.4926,
+      "step": 11028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4186190262731162e-05,
+      "loss": 0.4801,
+      "step": 11029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185172444320007e-05,
+      "loss": 0.4318,
+      "step": 11030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.418415457334285e-05,
+      "loss": 0.481,
+      "step": 11031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4183136649812477e-05,
+      "loss": 0.4448,
+      "step": 11032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182118673741673e-05,
+      "loss": 0.4658,
+      "step": 11033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181100645143224e-05,
+      "loss": 0.473,
+      "step": 11034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180082564029923e-05,
+      "loss": 0.485,
+      "step": 11035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179064430414545e-05,
+      "loss": 0.4594,
+      "step": 11036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4178046244309882e-05,
+      "loss": 0.4544,
+      "step": 11037
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177028005728728e-05,
+      "loss": 0.4854,
+      "step": 11038
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4176009714683865e-05,
+      "loss": 0.4704,
+      "step": 11039
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174991371188089e-05,
+      "loss": 0.4751,
+      "step": 11040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4173972975254186e-05,
+      "loss": 0.4669,
+      "step": 11041
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172954526894948e-05,
+      "loss": 0.4556,
+      "step": 11042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417193602612317e-05,
+      "loss": 0.4917,
+      "step": 11043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4170917472951639e-05,
+      "loss": 0.4343,
+      "step": 11044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169898867393155e-05,
+      "loss": 0.4669,
+      "step": 11045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168880209460505e-05,
+      "loss": 0.4712,
+      "step": 11046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167861499166486e-05,
+      "loss": 0.4795,
+      "step": 11047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166842736523892e-05,
+      "loss": 0.5185,
+      "step": 11048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165823921545521e-05,
+      "loss": 0.4907,
+      "step": 11049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4164805054244168e-05,
+      "loss": 0.4807,
+      "step": 11050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416378613463263e-05,
+      "loss": 0.4637,
+      "step": 11051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162767162723707e-05,
+      "loss": 0.4747,
+      "step": 11052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161748138530195e-05,
+      "loss": 0.4617,
+      "step": 11053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416072906206489e-05,
+      "loss": 0.4273,
+      "step": 11054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4159709933340598e-05,
+      "loss": 0.4419,
+      "step": 11055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158690752370117e-05,
+      "loss": 0.4909,
+      "step": 11056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4157671519166245e-05,
+      "loss": 0.4713,
+      "step": 11057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156652233741786e-05,
+      "loss": 0.4186,
+      "step": 11058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155632896109543e-05,
+      "loss": 0.4678,
+      "step": 11059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4154613506282315e-05,
+      "loss": 0.5381,
+      "step": 11060
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153594064272916e-05,
+      "loss": 0.4364,
+      "step": 11061
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152574570094136e-05,
+      "loss": 0.4518,
+      "step": 11062
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415155502375879e-05,
+      "loss": 0.4177,
+      "step": 11063
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415053542527968e-05,
+      "loss": 0.5276,
+      "step": 11064
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149515774669608e-05,
+      "loss": 0.4582,
+      "step": 11065
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4148496071941394e-05,
+      "loss": 0.4902,
+      "step": 11066
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147476317107833e-05,
+      "loss": 0.4866,
+      "step": 11067
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146456510181734e-05,
+      "loss": 0.473,
+      "step": 11068
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4145436651175913e-05,
+      "loss": 0.4943,
+      "step": 11069
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144416740103174e-05,
+      "loss": 0.4928,
+      "step": 11070
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143396776976331e-05,
+      "loss": 0.4343,
+      "step": 11071
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414237676180819e-05,
+      "loss": 0.4679,
+      "step": 11072
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141356694611565e-05,
+      "loss": 0.4935,
+      "step": 11073
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140336575399267e-05,
+      "loss": 0.4376,
+      "step": 11074
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139316404184111e-05,
+      "loss": 0.389,
+      "step": 11075
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413829618097891e-05,
+      "loss": 0.5149,
+      "step": 11076
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137275905796477e-05,
+      "loss": 0.5047,
+      "step": 11077
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4136255578649625e-05,
+      "loss": 0.4797,
+      "step": 11078
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135235199551171e-05,
+      "loss": 0.4225,
+      "step": 11079
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134214768513932e-05,
+      "loss": 0.4441,
+      "step": 11080
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4133194285550726e-05,
+      "loss": 0.4354,
+      "step": 11081
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132173750674367e-05,
+      "loss": 0.5054,
+      "step": 11082
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131153163897674e-05,
+      "loss": 0.4902,
+      "step": 11083
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4130132525233467e-05,
+      "loss": 0.4897,
+      "step": 11084
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4129111834694562e-05,
+      "loss": 0.4581,
+      "step": 11085
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128091092293784e-05,
+      "loss": 0.4229,
+      "step": 11086
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127070298043949e-05,
+      "loss": 0.4929,
+      "step": 11087
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126049451957879e-05,
+      "loss": 0.482,
+      "step": 11088
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4125028554048399e-05,
+      "loss": 0.5118,
+      "step": 11089
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412400760432833e-05,
+      "loss": 0.4654,
+      "step": 11090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4122986602810489e-05,
+      "loss": 0.4725,
+      "step": 11091
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412196554950771e-05,
+      "loss": 0.4401,
+      "step": 11092
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120944444432813e-05,
+      "loss": 0.4516,
+      "step": 11093
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119923287598622e-05,
+      "loss": 0.4589,
+      "step": 11094
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118902079017964e-05,
+      "loss": 0.5243,
+      "step": 11095
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117880818703664e-05,
+      "loss": 0.519,
+      "step": 11096
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411685950666855e-05,
+      "loss": 0.5049,
+      "step": 11097
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115838142925453e-05,
+      "loss": 0.444,
+      "step": 11098
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114816727487196e-05,
+      "loss": 0.4999,
+      "step": 11099
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113795260366612e-05,
+      "loss": 0.4504,
+      "step": 11100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112773741576526e-05,
+      "loss": 0.466,
+      "step": 11101
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4111752171129773e-05,
+      "loss": 0.4383,
+      "step": 11102
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110730549039185e-05,
+      "loss": 0.444,
+      "step": 11103
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410970887531759e-05,
+      "loss": 0.4676,
+      "step": 11104
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4108687149977817e-05,
+      "loss": 0.5088,
+      "step": 11105
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107665373032708e-05,
+      "loss": 0.4472,
+      "step": 11106
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4106643544495092e-05,
+      "loss": 0.4535,
+      "step": 11107
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105621664377802e-05,
+      "loss": 0.4762,
+      "step": 11108
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104599732693673e-05,
+      "loss": 0.5238,
+      "step": 11109
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4103577749455543e-05,
+      "loss": 0.4278,
+      "step": 11110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102555714676244e-05,
+      "loss": 0.4838,
+      "step": 11111
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101533628368618e-05,
+      "loss": 0.4797,
+      "step": 11112
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41005114905455e-05,
+      "loss": 0.4637,
+      "step": 11113
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099489301219727e-05,
+      "loss": 0.4595,
+      "step": 11114
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098467060404139e-05,
+      "loss": 0.5036,
+      "step": 11115
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4097444768111573e-05,
+      "loss": 0.4841,
+      "step": 11116
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096422424354875e-05,
+      "loss": 0.4358,
+      "step": 11117
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409540002914688e-05,
+      "loss": 0.5233,
+      "step": 11118
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094377582500432e-05,
+      "loss": 0.4731,
+      "step": 11119
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093355084428373e-05,
+      "loss": 0.4647,
+      "step": 11120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092332534943542e-05,
+      "loss": 0.5006,
+      "step": 11121
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091309934058786e-05,
+      "loss": 0.491,
+      "step": 11122
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090287281786949e-05,
+      "loss": 0.4963,
+      "step": 11123
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089264578140875e-05,
+      "loss": 0.508,
+      "step": 11124
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4088241823133408e-05,
+      "loss": 0.5232,
+      "step": 11125
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4087219016777397e-05,
+      "loss": 0.4874,
+      "step": 11126
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4086196159085686e-05,
+      "loss": 0.4682,
+      "step": 11127
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085173250071117e-05,
+      "loss": 0.4605,
+      "step": 11128
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4084150289746547e-05,
+      "loss": 0.501,
+      "step": 11129
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083127278124825e-05,
+      "loss": 0.451,
+      "step": 11130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4082104215218788e-05,
+      "loss": 0.4739,
+      "step": 11131
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4081081101041298e-05,
+      "loss": 0.4359,
+      "step": 11132
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080057935605199e-05,
+      "loss": 0.4877,
+      "step": 11133
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079034718923344e-05,
+      "loss": 0.4766,
+      "step": 11134
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4078011451008586e-05,
+      "loss": 0.4632,
+      "step": 11135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4076988131873775e-05,
+      "loss": 0.45,
+      "step": 11136
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4075964761531765e-05,
+      "loss": 0.4357,
+      "step": 11137
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074941339995407e-05,
+      "loss": 0.5191,
+      "step": 11138
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4073917867277557e-05,
+      "loss": 0.4731,
+      "step": 11139
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4072894343391074e-05,
+      "loss": 0.4891,
+      "step": 11140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071870768348812e-05,
+      "loss": 0.4715,
+      "step": 11141
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070847142163619e-05,
+      "loss": 0.4741,
+      "step": 11142
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069823464848363e-05,
+      "loss": 0.4792,
+      "step": 11143
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068799736415893e-05,
+      "loss": 0.4888,
+      "step": 11144
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067775956879071e-05,
+      "loss": 0.3979,
+      "step": 11145
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066752126250756e-05,
+      "loss": 0.4918,
+      "step": 11146
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4065728244543806e-05,
+      "loss": 0.4468,
+      "step": 11147
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406470431177108e-05,
+      "loss": 0.4603,
+      "step": 11148
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063680327945442e-05,
+      "loss": 0.4568,
+      "step": 11149
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062656293079752e-05,
+      "loss": 0.4475,
+      "step": 11150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406163220718687e-05,
+      "loss": 0.4462,
+      "step": 11151
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4060608070279663e-05,
+      "loss": 0.4362,
+      "step": 11152
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059583882370986e-05,
+      "loss": 0.4264,
+      "step": 11153
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058559643473712e-05,
+      "loss": 0.4749,
+      "step": 11154
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40575353536007e-05,
+      "loss": 0.4242,
+      "step": 11155
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056511012764817e-05,
+      "loss": 0.4701,
+      "step": 11156
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055486620978932e-05,
+      "loss": 0.3983,
+      "step": 11157
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4054462178255904e-05,
+      "loss": 0.5536,
+      "step": 11158
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4053437684608602e-05,
+      "loss": 0.4942,
+      "step": 11159
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052413140049898e-05,
+      "loss": 0.4481,
+      "step": 11160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051388544592662e-05,
+      "loss": 0.4645,
+      "step": 11161
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050363898249756e-05,
+      "loss": 0.4654,
+      "step": 11162
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049339201034049e-05,
+      "loss": 0.4214,
+      "step": 11163
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4048314452958418e-05,
+      "loss": 0.4418,
+      "step": 11164
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404728965403573e-05,
+      "loss": 0.4548,
+      "step": 11165
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046264804278857e-05,
+      "loss": 0.4421,
+      "step": 11166
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4045239903700675e-05,
+      "loss": 0.4817,
+      "step": 11167
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404421495231405e-05,
+      "loss": 0.5017,
+      "step": 11168
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043189950131859e-05,
+      "loss": 0.4642,
+      "step": 11169
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042164897166975e-05,
+      "loss": 0.4501,
+      "step": 11170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041139793432274e-05,
+      "loss": 0.5234,
+      "step": 11171
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040114638940635e-05,
+      "loss": 0.4643,
+      "step": 11172
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4039089433704924e-05,
+      "loss": 0.4796,
+      "step": 11173
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4038064177738026e-05,
+      "loss": 0.5039,
+      "step": 11174
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4037038871052815e-05,
+      "loss": 0.4508,
+      "step": 11175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036013513662168e-05,
+      "loss": 0.4878,
+      "step": 11176
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403498810557897e-05,
+      "loss": 0.4569,
+      "step": 11177
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033962646816093e-05,
+      "loss": 0.4382,
+      "step": 11178
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032937137386418e-05,
+      "loss": 0.4551,
+      "step": 11179
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031911577302827e-05,
+      "loss": 0.4595,
+      "step": 11180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030885966578202e-05,
+      "loss": 0.5478,
+      "step": 11181
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4029860305225424e-05,
+      "loss": 0.5092,
+      "step": 11182
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028834593257375e-05,
+      "loss": 0.5226,
+      "step": 11183
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027808830686936e-05,
+      "loss": 0.4627,
+      "step": 11184
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026783017526993e-05,
+      "loss": 0.4667,
+      "step": 11185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4025757153790429e-05,
+      "loss": 0.4562,
+      "step": 11186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402473123949013e-05,
+      "loss": 0.4227,
+      "step": 11187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023705274638984e-05,
+      "loss": 0.4155,
+      "step": 11188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022679259249871e-05,
+      "loss": 0.4884,
+      "step": 11189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4021653193335683e-05,
+      "loss": 0.5035,
+      "step": 11190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020627076909304e-05,
+      "loss": 0.3774,
+      "step": 11191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019600909983627e-05,
+      "loss": 0.4407,
+      "step": 11192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018574692571535e-05,
+      "loss": 0.4757,
+      "step": 11193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4017548424685918e-05,
+      "loss": 0.4689,
+      "step": 11194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016522106339671e-05,
+      "loss": 0.4664,
+      "step": 11195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401549573754568e-05,
+      "loss": 0.4355,
+      "step": 11196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014469318316836e-05,
+      "loss": 0.4675,
+      "step": 11197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013442848666039e-05,
+      "loss": 0.4648,
+      "step": 11198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4012416328606168e-05,
+      "loss": 0.5063,
+      "step": 11199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4011389758150127e-05,
+      "loss": 0.4316,
+      "step": 11200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010363137310803e-05,
+      "loss": 0.4474,
+      "step": 11201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009336466101094e-05,
+      "loss": 0.5062,
+      "step": 11202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40083097445339e-05,
+      "loss": 0.4631,
+      "step": 11203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007282972622105e-05,
+      "loss": 0.4814,
+      "step": 11204
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4006256150378612e-05,
+      "loss": 0.5117,
+      "step": 11205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005229277816319e-05,
+      "loss": 0.4514,
+      "step": 11206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004202354948119e-05,
+      "loss": 0.5106,
+      "step": 11207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4003175381786918e-05,
+      "loss": 0.4817,
+      "step": 11208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4002148358345609e-05,
+      "loss": 0.4237,
+      "step": 11209
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001121284637087e-05,
+      "loss": 0.4562,
+      "step": 11210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000094160674263e-05,
+      "loss": 0.4637,
+      "step": 11211
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999066986470032e-05,
+      "loss": 0.4838,
+      "step": 11212
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998039762037295e-05,
+      "loss": 0.4866,
+      "step": 11213
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997012487388954e-05,
+      "loss": 0.4296,
+      "step": 11214
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995985162537913e-05,
+      "loss": 0.452,
+      "step": 11215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3994957787497073e-05,
+      "loss": 0.4293,
+      "step": 11216
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399393036227934e-05,
+      "loss": 0.4402,
+      "step": 11217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399290288689762e-05,
+      "loss": 0.4933,
+      "step": 11218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991875361364815e-05,
+      "loss": 0.4305,
+      "step": 11219
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399084778569383e-05,
+      "loss": 0.4516,
+      "step": 11220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989820159897577e-05,
+      "loss": 0.4378,
+      "step": 11221
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3988792483988953e-05,
+      "loss": 0.4865,
+      "step": 11222
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398776475798088e-05,
+      "loss": 0.4688,
+      "step": 11223
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986736981886254e-05,
+      "loss": 0.5089,
+      "step": 11224
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985709155717988e-05,
+      "loss": 0.3987,
+      "step": 11225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3984681279488993e-05,
+      "loss": 0.4789,
+      "step": 11226
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3983653353212176e-05,
+      "loss": 0.4747,
+      "step": 11227
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3982625376900451e-05,
+      "loss": 0.4966,
+      "step": 11228
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981597350566729e-05,
+      "loss": 0.4978,
+      "step": 11229
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980569274223921e-05,
+      "loss": 0.4867,
+      "step": 11230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397954114788494e-05,
+      "loss": 0.4311,
+      "step": 11231
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3978512971562699e-05,
+      "loss": 0.4136,
+      "step": 11232
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977484745270112e-05,
+      "loss": 0.4086,
+      "step": 11233
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976456469020092e-05,
+      "loss": 0.452,
+      "step": 11234
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975428142825562e-05,
+      "loss": 0.4521,
+      "step": 11235
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974399766699426e-05,
+      "loss": 0.4626,
+      "step": 11236
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397337134065461e-05,
+      "loss": 0.4562,
+      "step": 11237
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972342864704025e-05,
+      "loss": 0.4576,
+      "step": 11238
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397131433886059e-05,
+      "loss": 0.516,
+      "step": 11239
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970285763137229e-05,
+      "loss": 0.4386,
+      "step": 11240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969257137546854e-05,
+      "loss": 0.4284,
+      "step": 11241
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968228462102386e-05,
+      "loss": 0.5389,
+      "step": 11242
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967199736816745e-05,
+      "loss": 0.4717,
+      "step": 11243
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966170961702856e-05,
+      "loss": 0.4634,
+      "step": 11244
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396514213677364e-05,
+      "loss": 0.4967,
+      "step": 11245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964113262042012e-05,
+      "loss": 0.4749,
+      "step": 11246
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39630843375209e-05,
+      "loss": 0.4917,
+      "step": 11247
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962055363223228e-05,
+      "loss": 0.4355,
+      "step": 11248
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961026339161918e-05,
+      "loss": 0.4538,
+      "step": 11249
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959997265349898e-05,
+      "loss": 0.4408,
+      "step": 11250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958968141800089e-05,
+      "loss": 0.4459,
+      "step": 11251
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957938968525417e-05,
+      "loss": 0.4498,
+      "step": 11252
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395690974553881e-05,
+      "loss": 0.4188,
+      "step": 11253
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3955880472853193e-05,
+      "loss": 0.5147,
+      "step": 11254
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39548511504815e-05,
+      "loss": 0.4336,
+      "step": 11255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953821778436654e-05,
+      "loss": 0.4247,
+      "step": 11256
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3952792356731586e-05,
+      "loss": 0.4674,
+      "step": 11257
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3951762885379223e-05,
+      "loss": 0.4838,
+      "step": 11258
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950733364392495e-05,
+      "loss": 0.4537,
+      "step": 11259
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949703793784337e-05,
+      "loss": 0.4506,
+      "step": 11260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948674173567677e-05,
+      "loss": 0.4569,
+      "step": 11261
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947644503755451e-05,
+      "loss": 0.4187,
+      "step": 11262
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3946614784360586e-05,
+      "loss": 0.4286,
+      "step": 11263
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945585015396018e-05,
+      "loss": 0.4446,
+      "step": 11264
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944555196874683e-05,
+      "loss": 0.5205,
+      "step": 11265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943525328809513e-05,
+      "loss": 0.4955,
+      "step": 11266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942495411213446e-05,
+      "loss": 0.458,
+      "step": 11267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941465444099413e-05,
+      "loss": 0.4684,
+      "step": 11268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940435427480354e-05,
+      "loss": 0.4945,
+      "step": 11269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939405361369205e-05,
+      "loss": 0.4766,
+      "step": 11270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3938375245778905e-05,
+      "loss": 0.479,
+      "step": 11271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937345080722393e-05,
+      "loss": 0.4466,
+      "step": 11272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936314866212603e-05,
+      "loss": 0.4726,
+      "step": 11273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393528460226248e-05,
+      "loss": 0.4691,
+      "step": 11274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934254288884961e-05,
+      "loss": 0.526,
+      "step": 11275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3933223926092985e-05,
+      "loss": 0.4715,
+      "step": 11276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3932193513899503e-05,
+      "loss": 0.4697,
+      "step": 11277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931163052317448e-05,
+      "loss": 0.4793,
+      "step": 11278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393013254135976e-05,
+      "loss": 0.4837,
+      "step": 11279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392910198103939e-05,
+      "loss": 0.4353,
+      "step": 11280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928071371369279e-05,
+      "loss": 0.4277,
+      "step": 11281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927040712362374e-05,
+      "loss": 0.4445,
+      "step": 11282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926010004031617e-05,
+      "loss": 0.478,
+      "step": 11283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924979246389949e-05,
+      "loss": 0.4704,
+      "step": 11284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3923948439450327e-05,
+      "loss": 0.4512,
+      "step": 11285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922917583225693e-05,
+      "loss": 0.5122,
+      "step": 11286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921886677728992e-05,
+      "loss": 0.4451,
+      "step": 11287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920855722973179e-05,
+      "loss": 0.4709,
+      "step": 11288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3919824718971195e-05,
+      "loss": 0.5061,
+      "step": 11289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918793665735993e-05,
+      "loss": 0.4776,
+      "step": 11290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917762563280524e-05,
+      "loss": 0.5018,
+      "step": 11291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3916731411617736e-05,
+      "loss": 0.4568,
+      "step": 11292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915700210760586e-05,
+      "loss": 0.5315,
+      "step": 11293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914668960722018e-05,
+      "loss": 0.4666,
+      "step": 11294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913637661514991e-05,
+      "loss": 0.4953,
+      "step": 11295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912606313152452e-05,
+      "loss": 0.5176,
+      "step": 11296
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3911574915647364e-05,
+      "loss": 0.445,
+      "step": 11297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3910543469012674e-05,
+      "loss": 0.5357,
+      "step": 11298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909511973261339e-05,
+      "loss": 0.4667,
+      "step": 11299
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908480428406313e-05,
+      "loss": 0.5048,
+      "step": 11300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907448834460554e-05,
+      "loss": 0.4435,
+      "step": 11301
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906417191437021e-05,
+      "loss": 0.5132,
+      "step": 11302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905385499348667e-05,
+      "loss": 0.4943,
+      "step": 11303
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904353758208456e-05,
+      "loss": 0.4844,
+      "step": 11304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3903321968029339e-05,
+      "loss": 0.4474,
+      "step": 11305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902290128824282e-05,
+      "loss": 0.4766,
+      "step": 11306
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3901258240606241e-05,
+      "loss": 0.4287,
+      "step": 11307
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900226303388178e-05,
+      "loss": 0.4935,
+      "step": 11308
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899194317183059e-05,
+      "loss": 0.4706,
+      "step": 11309
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898162282003832e-05,
+      "loss": 0.4988,
+      "step": 11310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897130197863476e-05,
+      "loss": 0.4476,
+      "step": 11311
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896098064774944e-05,
+      "loss": 0.4475,
+      "step": 11312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.38950658827512e-05,
+      "loss": 0.4457,
+      "step": 11313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894033651805214e-05,
+      "loss": 0.4867,
+      "step": 11314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893001371949946e-05,
+      "loss": 0.4258,
+      "step": 11315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389196904319836e-05,
+      "loss": 0.4537,
+      "step": 11316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3890936665563427e-05,
+      "loss": 0.4601,
+      "step": 11317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889904239058112e-05,
+      "loss": 0.4459,
+      "step": 11318
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888871763695382e-05,
+      "loss": 0.4007,
+      "step": 11319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3887839239488206e-05,
+      "loss": 0.452,
+      "step": 11320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886806666449547e-05,
+      "loss": 0.4848,
+      "step": 11321
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3885774044592382e-05,
+      "loss": 0.5038,
+      "step": 11322
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3884741373929677e-05,
+      "loss": 0.4389,
+      "step": 11323
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883708654474402e-05,
+      "loss": 0.3757,
+      "step": 11324
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882675886239529e-05,
+      "loss": 0.4917,
+      "step": 11325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388164306923803e-05,
+      "loss": 0.4477,
+      "step": 11326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880610203482875e-05,
+      "loss": 0.4767,
+      "step": 11327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387957728898704e-05,
+      "loss": 0.4323,
+      "step": 11328
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38785443257635e-05,
+      "loss": 0.5046,
+      "step": 11329
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387751131382522e-05,
+      "loss": 0.491,
+      "step": 11330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876478253185183e-05,
+      "loss": 0.4169,
+      "step": 11331
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875445143856365e-05,
+      "loss": 0.4351,
+      "step": 11332
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3874411985851734e-05,
+      "loss": 0.4724,
+      "step": 11333
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873378779184275e-05,
+      "loss": 0.4926,
+      "step": 11334
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872345523866961e-05,
+      "loss": 0.5311,
+      "step": 11335
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3871312219912772e-05,
+      "loss": 0.5235,
+      "step": 11336
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870278867334682e-05,
+      "loss": 0.4798,
+      "step": 11337
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386924546614567e-05,
+      "loss": 0.4743,
+      "step": 11338
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3868212016358722e-05,
+      "loss": 0.4445,
+      "step": 11339
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867178517986817e-05,
+      "loss": 0.4612,
+      "step": 11340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3866144971042927e-05,
+      "loss": 0.5027,
+      "step": 11341
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865111375540044e-05,
+      "loss": 0.4156,
+      "step": 11342
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864077731491145e-05,
+      "loss": 0.4252,
+      "step": 11343
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863044038909212e-05,
+      "loss": 0.5279,
+      "step": 11344
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862010297807231e-05,
+      "loss": 0.4487,
+      "step": 11345
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860976508198185e-05,
+      "loss": 0.4855,
+      "step": 11346
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3859942670095055e-05,
+      "loss": 0.4865,
+      "step": 11347
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858908783510832e-05,
+      "loss": 0.5175,
+      "step": 11348
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857874848458497e-05,
+      "loss": 0.4824,
+      "step": 11349
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856840864951038e-05,
+      "loss": 0.4603,
+      "step": 11350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855806833001442e-05,
+      "loss": 0.5004,
+      "step": 11351
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854772752622694e-05,
+      "loss": 0.4796,
+      "step": 11352
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3853738623827787e-05,
+      "loss": 0.4185,
+      "step": 11353
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852704446629705e-05,
+      "loss": 0.5182,
+      "step": 11354
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385167022104144e-05,
+      "loss": 0.5005,
+      "step": 11355
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850635947075983e-05,
+      "loss": 0.4449,
+      "step": 11356
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849601624746321e-05,
+      "loss": 0.4698,
+      "step": 11357
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848567254065448e-05,
+      "loss": 0.4913,
+      "step": 11358
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3847532835046354e-05,
+      "loss": 0.5125,
+      "step": 11359
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846498367702032e-05,
+      "loss": 0.4747,
+      "step": 11360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3845463852045475e-05,
+      "loss": 0.4919,
+      "step": 11361
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844429288089678e-05,
+      "loss": 0.4446,
+      "step": 11362
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3843394675847635e-05,
+      "loss": 0.4028,
+      "step": 11363
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3842360015332335e-05,
+      "loss": 0.4378,
+      "step": 11364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841325306556781e-05,
+      "loss": 0.4374,
+      "step": 11365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840290549533966e-05,
+      "loss": 0.4492,
+      "step": 11366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839255744276887e-05,
+      "loss": 0.4527,
+      "step": 11367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3838220890798539e-05,
+      "loss": 0.509,
+      "step": 11368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837185989111924e-05,
+      "loss": 0.4872,
+      "step": 11369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836151039230035e-05,
+      "loss": 0.4058,
+      "step": 11370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3835116041165877e-05,
+      "loss": 0.4985,
+      "step": 11371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834080994932446e-05,
+      "loss": 0.4447,
+      "step": 11372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833045900542744e-05,
+      "loss": 0.4751,
+      "step": 11373
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832010758009769e-05,
+      "loss": 0.5057,
+      "step": 11374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3830975567346522e-05,
+      "loss": 0.4637,
+      "step": 11375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829940328566012e-05,
+      "loss": 0.4745,
+      "step": 11376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828905041681235e-05,
+      "loss": 0.4996,
+      "step": 11377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827869706705194e-05,
+      "loss": 0.458,
+      "step": 11378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.4646,
+      "step": 11379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825798892531351e-05,
+      "loss": 0.4934,
+      "step": 11380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382476341335955e-05,
+      "loss": 0.4449,
+      "step": 11381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382372788614851e-05,
+      "loss": 0.4599,
+      "step": 11382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822692310911231e-05,
+      "loss": 0.4801,
+      "step": 11383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821656687660725e-05,
+      "loss": 0.4775,
+      "step": 11384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3820621016409996e-05,
+      "loss": 0.4581,
+      "step": 11385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3819585297172053e-05,
+      "loss": 0.5308,
+      "step": 11386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818549529959906e-05,
+      "loss": 0.4608,
+      "step": 11387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381751371478656e-05,
+      "loss": 0.5259,
+      "step": 11388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381647785166503e-05,
+      "loss": 0.4736,
+      "step": 11389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815441940608323e-05,
+      "loss": 0.5081,
+      "step": 11390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814405981629455e-05,
+      "loss": 0.4529,
+      "step": 11391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381336997474143e-05,
+      "loss": 0.4339,
+      "step": 11392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812333919957268e-05,
+      "loss": 0.4453,
+      "step": 11393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811297817289978e-05,
+      "loss": 0.4818,
+      "step": 11394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810261666752571e-05,
+      "loss": 0.4358,
+      "step": 11395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809225468358066e-05,
+      "loss": 0.5109,
+      "step": 11396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3808189222119475e-05,
+      "loss": 0.4314,
+      "step": 11397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807152928049817e-05,
+      "loss": 0.4493,
+      "step": 11398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3806116586162102e-05,
+      "loss": 0.4538,
+      "step": 11399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805080196469351e-05,
+      "loss": 0.562,
+      "step": 11400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804043758984578e-05,
+      "loss": 0.452,
+      "step": 11401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803007273720802e-05,
+      "loss": 0.5439,
+      "step": 11402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801970740691045e-05,
+      "loss": 0.4157,
+      "step": 11403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380093415990832e-05,
+      "loss": 0.4942,
+      "step": 11404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799897531385648e-05,
+      "loss": 0.4319,
+      "step": 11405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3798860855136054e-05,
+      "loss": 0.4993,
+      "step": 11406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379782413117255e-05,
+      "loss": 0.3894,
+      "step": 11407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796787359508167e-05,
+      "loss": 0.4385,
+      "step": 11408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795750540155918e-05,
+      "loss": 0.4988,
+      "step": 11409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3794713673128832e-05,
+      "loss": 0.3949,
+      "step": 11410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379367675843993e-05,
+      "loss": 0.4661,
+      "step": 11411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3792639796102234e-05,
+      "loss": 0.461,
+      "step": 11412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379160278612877e-05,
+      "loss": 0.5125,
+      "step": 11413
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3790565728532566e-05,
+      "loss": 0.4881,
+      "step": 11414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.378952862332664e-05,
+      "loss": 0.4982,
+      "step": 11415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3788491470524023e-05,
+      "loss": 0.5014,
+      "step": 11416
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3787454270137742e-05,
+      "loss": 0.4785,
+      "step": 11417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378641702218082e-05,
+      "loss": 0.4698,
+      "step": 11418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3785379726666295e-05,
+      "loss": 0.4577,
+      "step": 11419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3784342383607182e-05,
+      "loss": 0.4932,
+      "step": 11420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378330499301652e-05,
+      "loss": 0.4713,
+      "step": 11421
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782267554907333e-05,
+      "loss": 0.4479,
+      "step": 11422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781230069292657e-05,
+      "loss": 0.4442,
+      "step": 11423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780192536185519e-05,
+      "loss": 0.4472,
+      "step": 11424
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377915495559895e-05,
+      "loss": 0.4356,
+      "step": 11425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778117327545981e-05,
+      "loss": 0.4561,
+      "step": 11426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777079652039649e-05,
+      "loss": 0.4456,
+      "step": 11427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776041929092986e-05,
+      "loss": 0.4824,
+      "step": 11428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3775004158719023e-05,
+      "loss": 0.4591,
+      "step": 11429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773966340930798e-05,
+      "loss": 0.4768,
+      "step": 11430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772928475741344e-05,
+      "loss": 0.3897,
+      "step": 11431
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771890563163694e-05,
+      "loss": 0.4729,
+      "step": 11432
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377085260321089e-05,
+      "loss": 0.4704,
+      "step": 11433
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3769814595895967e-05,
+      "loss": 0.4613,
+      "step": 11434
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768776541231961e-05,
+      "loss": 0.5187,
+      "step": 11435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376773843923191e-05,
+      "loss": 0.4459,
+      "step": 11436
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766700289908854e-05,
+      "loss": 0.4396,
+      "step": 11437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376566209327583e-05,
+      "loss": 0.4736,
+      "step": 11438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3764623849345879e-05,
+      "loss": 0.4083,
+      "step": 11439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763585558132042e-05,
+      "loss": 0.4546,
+      "step": 11440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376254721964736e-05,
+      "loss": 0.4743,
+      "step": 11441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3761508833904876e-05,
+      "loss": 0.4012,
+      "step": 11442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760470400917627e-05,
+      "loss": 0.4637,
+      "step": 11443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759431920698661e-05,
+      "loss": 0.4231,
+      "step": 11444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375839339326102e-05,
+      "loss": 0.52,
+      "step": 11445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757354818617745e-05,
+      "loss": 0.4839,
+      "step": 11446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3756316196781887e-05,
+      "loss": 0.4698,
+      "step": 11447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3755277527766484e-05,
+      "loss": 0.5083,
+      "step": 11448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754238811584583e-05,
+      "loss": 0.4358,
+      "step": 11449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375320004824924e-05,
+      "loss": 0.491,
+      "step": 11450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3752161237773489e-05,
+      "loss": 0.4504,
+      "step": 11451
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375112238017038e-05,
+      "loss": 0.4772,
+      "step": 11452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750083475452969e-05,
+      "loss": 0.5291,
+      "step": 11453
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749044523634298e-05,
+      "loss": 0.4699,
+      "step": 11454
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748005524727417e-05,
+      "loss": 0.4579,
+      "step": 11455
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746966478745378e-05,
+      "loss": 0.4942,
+      "step": 11456
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3745927385701227e-05,
+      "loss": 0.4579,
+      "step": 11457
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374488824560802e-05,
+      "loss": 0.475,
+      "step": 11458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3743849058478809e-05,
+      "loss": 0.4645,
+      "step": 11459
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3742809824326641e-05,
+      "loss": 0.4774,
+      "step": 11460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741770543164574e-05,
+      "loss": 0.5104,
+      "step": 11461
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740731215005658e-05,
+      "loss": 0.4553,
+      "step": 11462
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373969183986295e-05,
+      "loss": 0.4488,
+      "step": 11463
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37386524177495e-05,
+      "loss": 0.4589,
+      "step": 11464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373761294867837e-05,
+      "loss": 0.4677,
+      "step": 11465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373657343266261e-05,
+      "loss": 0.4448,
+      "step": 11466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735533869715281e-05,
+      "loss": 0.4683,
+      "step": 11467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3734494259849436e-05,
+      "loss": 0.5005,
+      "step": 11468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733454603078132e-05,
+      "loss": 0.4914,
+      "step": 11469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732414899414428e-05,
+      "loss": 0.4427,
+      "step": 11470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731375148871387e-05,
+      "loss": 0.4465,
+      "step": 11471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3730335351462065e-05,
+      "loss": 0.4994,
+      "step": 11472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729295507199523e-05,
+      "loss": 0.4384,
+      "step": 11473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728255616096818e-05,
+      "loss": 0.4605,
+      "step": 11474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727215678167015e-05,
+      "loss": 0.4833,
+      "step": 11475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3726175693423174e-05,
+      "loss": 0.4332,
+      "step": 11476
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372513566187836e-05,
+      "loss": 0.4831,
+      "step": 11477
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3724095583545628e-05,
+      "loss": 0.4829,
+      "step": 11478
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372305545843805e-05,
+      "loss": 0.4992,
+      "step": 11479
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3722015286568686e-05,
+      "loss": 0.468,
+      "step": 11480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3720975067950602e-05,
+      "loss": 0.5363,
+      "step": 11481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719934802596866e-05,
+      "loss": 0.4467,
+      "step": 11482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3718894490520536e-05,
+      "loss": 0.4839,
+      "step": 11483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717854131734684e-05,
+      "loss": 0.4439,
+      "step": 11484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716813726252377e-05,
+      "loss": 0.4809,
+      "step": 11485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371577327408668e-05,
+      "loss": 0.4207,
+      "step": 11486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714732775250666e-05,
+      "loss": 0.4835,
+      "step": 11487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713692229757398e-05,
+      "loss": 0.4162,
+      "step": 11488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712651637619944e-05,
+      "loss": 0.4129,
+      "step": 11489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3711610998851384e-05,
+      "loss": 0.4668,
+      "step": 11490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710570313464778e-05,
+      "loss": 0.4207,
+      "step": 11491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709529581473203e-05,
+      "loss": 0.4683,
+      "step": 11492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708488802889725e-05,
+      "loss": 0.479,
+      "step": 11493
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707447977727424e-05,
+      "loss": 0.4017,
+      "step": 11494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706407105999367e-05,
+      "loss": 0.4414,
+      "step": 11495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370536618771863e-05,
+      "loss": 0.5109,
+      "step": 11496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704325222898284e-05,
+      "loss": 0.5194,
+      "step": 11497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370328421155141e-05,
+      "loss": 0.4704,
+      "step": 11498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702243153691076e-05,
+      "loss": 0.4771,
+      "step": 11499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370120204933036e-05,
+      "loss": 0.5091,
+      "step": 11500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370016089848234e-05,
+      "loss": 0.492,
+      "step": 11501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369911970116009e-05,
+      "loss": 0.4699,
+      "step": 11502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698078457376692e-05,
+      "loss": 0.4263,
+      "step": 11503
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697037167145222e-05,
+      "loss": 0.443,
+      "step": 11504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695995830478756e-05,
+      "loss": 0.5028,
+      "step": 11505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3694954447390374e-05,
+      "loss": 0.513,
+      "step": 11506
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3693913017893161e-05,
+      "loss": 0.5142,
+      "step": 11507
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692871542000192e-05,
+      "loss": 0.4867,
+      "step": 11508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369183001972455e-05,
+      "loss": 0.4742,
+      "step": 11509
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3690788451079315e-05,
+      "loss": 0.5049,
+      "step": 11510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3689746836077573e-05,
+      "loss": 0.47,
+      "step": 11511
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3688705174732399e-05,
+      "loss": 0.42,
+      "step": 11512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687663467056886e-05,
+      "loss": 0.4782,
+      "step": 11513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3686621713064115e-05,
+      "loss": 0.4805,
+      "step": 11514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685579912767162e-05,
+      "loss": 0.4797,
+      "step": 11515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684538066179123e-05,
+      "loss": 0.4519,
+      "step": 11516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3683496173313078e-05,
+      "loss": 0.436,
+      "step": 11517
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682454234182117e-05,
+      "loss": 0.5044,
+      "step": 11518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3681412248799327e-05,
+      "loss": 0.4687,
+      "step": 11519
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680370217177786e-05,
+      "loss": 0.4408,
+      "step": 11520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679328139330592e-05,
+      "loss": 0.4761,
+      "step": 11521
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678286015270833e-05,
+      "loss": 0.4815,
+      "step": 11522
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677243845011589e-05,
+      "loss": 0.428,
+      "step": 11523
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3676201628565965e-05,
+      "loss": 0.4475,
+      "step": 11524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675159365947038e-05,
+      "loss": 0.4177,
+      "step": 11525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674117057167902e-05,
+      "loss": 0.4946,
+      "step": 11526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673074702241654e-05,
+      "loss": 0.4133,
+      "step": 11527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367203230118138e-05,
+      "loss": 0.4937,
+      "step": 11528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3670989854000174e-05,
+      "loss": 0.447,
+      "step": 11529
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669947360711132e-05,
+      "loss": 0.5306,
+      "step": 11530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668904821327345e-05,
+      "loss": 0.4684,
+      "step": 11531
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667862235861904e-05,
+      "loss": 0.5193,
+      "step": 11532
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666819604327912e-05,
+      "loss": 0.4852,
+      "step": 11533
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366577692673846e-05,
+      "loss": 0.5055,
+      "step": 11534
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664734203106645e-05,
+      "loss": 0.424,
+      "step": 11535
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663691433445561e-05,
+      "loss": 0.4662,
+      "step": 11536
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3662648617768306e-05,
+      "loss": 0.5172,
+      "step": 11537
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661605756087984e-05,
+      "loss": 0.5179,
+      "step": 11538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660562848417684e-05,
+      "loss": 0.4759,
+      "step": 11539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659519894770515e-05,
+      "loss": 0.4952,
+      "step": 11540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3658476895159568e-05,
+      "loss": 0.467,
+      "step": 11541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657433849597945e-05,
+      "loss": 0.538,
+      "step": 11542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3656390758098748e-05,
+      "loss": 0.5166,
+      "step": 11543
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655347620675081e-05,
+      "loss": 0.4434,
+      "step": 11544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654304437340043e-05,
+      "loss": 0.5307,
+      "step": 11545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653261208106736e-05,
+      "loss": 0.5243,
+      "step": 11546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3652217932988263e-05,
+      "loss": 0.4608,
+      "step": 11547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651174611997728e-05,
+      "loss": 0.437,
+      "step": 11548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3650131245148233e-05,
+      "loss": 0.4978,
+      "step": 11549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364908783245289e-05,
+      "loss": 0.4255,
+      "step": 11550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648044373924798e-05,
+      "loss": 0.4818,
+      "step": 11551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647000869577063e-05,
+      "loss": 0.4526,
+      "step": 11552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3645957319422794e-05,
+      "loss": 0.4797,
+      "step": 11553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644913723475095e-05,
+      "loss": 0.3899,
+      "step": 11554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643870081747076e-05,
+      "loss": 0.4133,
+      "step": 11555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3642826394251846e-05,
+      "loss": 0.5079,
+      "step": 11556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641782661002512e-05,
+      "loss": 0.4736,
+      "step": 11557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640738882012183e-05,
+      "loss": 0.4959,
+      "step": 11558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3639695057293971e-05,
+      "loss": 0.4702,
+      "step": 11559
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3638651186860985e-05,
+      "loss": 0.4958,
+      "step": 11560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637607270726338e-05,
+      "loss": 0.4272,
+      "step": 11561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636563308903139e-05,
+      "loss": 0.4579,
+      "step": 11562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635519301404498e-05,
+      "loss": 0.4739,
+      "step": 11563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634475248243533e-05,
+      "loss": 0.4874,
+      "step": 11564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633431149433356e-05,
+      "loss": 0.5345,
+      "step": 11565
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363238700498708e-05,
+      "loss": 0.4802,
+      "step": 11566
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631342814917821e-05,
+      "loss": 0.4392,
+      "step": 11567
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363029857923869e-05,
+      "loss": 0.4276,
+      "step": 11568
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629254297962807e-05,
+      "loss": 0.463,
+      "step": 11569
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3628209971103285e-05,
+      "loss": 0.4723,
+      "step": 11570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627165598673246e-05,
+      "loss": 0.4963,
+      "step": 11571
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626121180685803e-05,
+      "loss": 0.4892,
+      "step": 11572
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3625076717154074e-05,
+      "loss": 0.4772,
+      "step": 11573
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624032208091177e-05,
+      "loss": 0.4645,
+      "step": 11574
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622987653510233e-05,
+      "loss": 0.4546,
+      "step": 11575
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3621943053424362e-05,
+      "loss": 0.4906,
+      "step": 11576
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620898407846686e-05,
+      "loss": 0.4663,
+      "step": 11577
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3619853716790318e-05,
+      "loss": 0.3996,
+      "step": 11578
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361880898026839e-05,
+      "loss": 0.462,
+      "step": 11579
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3617764198294015e-05,
+      "loss": 0.4864,
+      "step": 11580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361671937088032e-05,
+      "loss": 0.4844,
+      "step": 11581
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361567449804043e-05,
+      "loss": 0.4464,
+      "step": 11582
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361462957978746e-05,
+      "loss": 0.4639,
+      "step": 11583
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613584616134543e-05,
+      "loss": 0.4598,
+      "step": 11584
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36125396070948e-05,
+      "loss": 0.4588,
+      "step": 11585
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3611494552681356e-05,
+      "loss": 0.4802,
+      "step": 11586
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3610449452907345e-05,
+      "loss": 0.476,
+      "step": 11587
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360940430778588e-05,
+      "loss": 0.4768,
+      "step": 11588
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608359117330094e-05,
+      "loss": 0.4931,
+      "step": 11589
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360731388155312e-05,
+      "loss": 0.4741,
+      "step": 11590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3606268600468078e-05,
+      "loss": 0.4563,
+      "step": 11591
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3605223274088104e-05,
+      "loss": 0.4793,
+      "step": 11592
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604177902426323e-05,
+      "loss": 0.4859,
+      "step": 11593
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3603132485495863e-05,
+      "loss": 0.4522,
+      "step": 11594
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602087023309858e-05,
+      "loss": 0.489,
+      "step": 11595
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601041515881438e-05,
+      "loss": 0.4851,
+      "step": 11596
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599995963223736e-05,
+      "loss": 0.4542,
+      "step": 11597
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598950365349884e-05,
+      "loss": 0.4002,
+      "step": 11598
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597904722273012e-05,
+      "loss": 0.4079,
+      "step": 11599
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596859034006255e-05,
+      "loss": 0.4506,
+      "step": 11600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3595813300562748e-05,
+      "loss": 0.4637,
+      "step": 11601
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594767521955626e-05,
+      "loss": 0.4204,
+      "step": 11602
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593721698198019e-05,
+      "loss": 0.4594,
+      "step": 11603
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359267582930307e-05,
+      "loss": 0.5023,
+      "step": 11604
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359162991528391e-05,
+      "loss": 0.4855,
+      "step": 11605
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590583956153673e-05,
+      "loss": 0.4993,
+      "step": 11606
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3589537951925507e-05,
+      "loss": 0.4565,
+      "step": 11607
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358849190261254e-05,
+      "loss": 0.4945,
+      "step": 11608
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587445808227914e-05,
+      "loss": 0.4632,
+      "step": 11609
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586399668784769e-05,
+      "loss": 0.4921,
+      "step": 11610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358535348429624e-05,
+      "loss": 0.5133,
+      "step": 11611
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584307254775473e-05,
+      "loss": 0.5584,
+      "step": 11612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583260980235607e-05,
+      "loss": 0.469,
+      "step": 11613
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3582214660689782e-05,
+      "loss": 0.4819,
+      "step": 11614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581168296151138e-05,
+      "loss": 0.496,
+      "step": 11615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580121886632821e-05,
+      "loss": 0.488,
+      "step": 11616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579075432147972e-05,
+      "loss": 0.4432,
+      "step": 11617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578028932709735e-05,
+      "loss": 0.4614,
+      "step": 11618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576982388331258e-05,
+      "loss": 0.4773,
+      "step": 11619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3575935799025677e-05,
+      "loss": 0.4713,
+      "step": 11620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574889164806144e-05,
+      "loss": 0.488,
+      "step": 11621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573842485685802e-05,
+      "loss": 0.4483,
+      "step": 11622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35727957616778e-05,
+      "loss": 0.4015,
+      "step": 11623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571748992795283e-05,
+      "loss": 0.4989,
+      "step": 11624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570702179051399e-05,
+      "loss": 0.4564,
+      "step": 11625
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569655320459291e-05,
+      "loss": 0.4573,
+      "step": 11626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568608417032116e-05,
+      "loss": 0.4697,
+      "step": 11627
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356756146878302e-05,
+      "loss": 0.4499,
+      "step": 11628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3566514475725151e-05,
+      "loss": 0.4509,
+      "step": 11629
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565467437871662e-05,
+      "loss": 0.4764,
+      "step": 11630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3564420355235699e-05,
+      "loss": 0.4481,
+      "step": 11631
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563373227830418e-05,
+      "loss": 0.4401,
+      "step": 11632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356232605566897e-05,
+      "loss": 0.3993,
+      "step": 11633
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356127883876451e-05,
+      "loss": 0.4817,
+      "step": 11634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560231577130185e-05,
+      "loss": 0.4724,
+      "step": 11635
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559184270779154e-05,
+      "loss": 0.4394,
+      "step": 11636
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3558136919724568e-05,
+      "loss": 0.438,
+      "step": 11637
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557089523979583e-05,
+      "loss": 0.5033,
+      "step": 11638
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3556042083557354e-05,
+      "loss": 0.4779,
+      "step": 11639
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554994598471036e-05,
+      "loss": 0.4378,
+      "step": 11640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355394706873379e-05,
+      "loss": 0.4581,
+      "step": 11641
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355289949435877e-05,
+      "loss": 0.4567,
+      "step": 11642
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551851875359127e-05,
+      "loss": 0.4028,
+      "step": 11643
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3550804211748032e-05,
+      "loss": 0.4827,
+      "step": 11644
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549756503538636e-05,
+      "loss": 0.4952,
+      "step": 11645
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35487087507441e-05,
+      "loss": 0.524,
+      "step": 11646
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547660953377581e-05,
+      "loss": 0.4493,
+      "step": 11647
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546613111452242e-05,
+      "loss": 0.4817,
+      "step": 11648
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545565224981245e-05,
+      "loss": 0.4967,
+      "step": 11649
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354451729397775e-05,
+      "loss": 0.4631,
+      "step": 11650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354346931845492e-05,
+      "loss": 0.5,
+      "step": 11651
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542421298425916e-05,
+      "loss": 0.4645,
+      "step": 11652
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541373233903903e-05,
+      "loss": 0.5124,
+      "step": 11653
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354032512490204e-05,
+      "loss": 0.4807,
+      "step": 11654
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539276971433498e-05,
+      "loss": 0.4509,
+      "step": 11655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3538228773511443e-05,
+      "loss": 0.4356,
+      "step": 11656
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353718053114903e-05,
+      "loss": 0.4913,
+      "step": 11657
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3536132244359434e-05,
+      "loss": 0.5231,
+      "step": 11658
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535083913155819e-05,
+      "loss": 0.4883,
+      "step": 11659
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353403553755135e-05,
+      "loss": 0.4209,
+      "step": 11660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3532987117559201e-05,
+      "loss": 0.4567,
+      "step": 11661
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531938653192533e-05,
+      "loss": 0.4655,
+      "step": 11662
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3530890144464519e-05,
+      "loss": 0.4782,
+      "step": 11663
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529841591388327e-05,
+      "loss": 0.4735,
+      "step": 11664
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528792993977124e-05,
+      "loss": 0.4384,
+      "step": 11665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3527744352244087e-05,
+      "loss": 0.4383,
+      "step": 11666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526695666202386e-05,
+      "loss": 0.5279,
+      "step": 11667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525646935865183e-05,
+      "loss": 0.4496,
+      "step": 11668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352459816124566e-05,
+      "loss": 0.4201,
+      "step": 11669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523549342356988e-05,
+      "loss": 0.4208,
+      "step": 11670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522500479212337e-05,
+      "loss": 0.4472,
+      "step": 11671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521451571824884e-05,
+      "loss": 0.5061,
+      "step": 11672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520402620207803e-05,
+      "loss": 0.4715,
+      "step": 11673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3519353624374263e-05,
+      "loss": 0.4709,
+      "step": 11674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3518304584337447e-05,
+      "loss": 0.4679,
+      "step": 11675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517255500110531e-05,
+      "loss": 0.4402,
+      "step": 11676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516206371706686e-05,
+      "loss": 0.4758,
+      "step": 11677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515157199139096e-05,
+      "loss": 0.4504,
+      "step": 11678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514107982420929e-05,
+      "loss": 0.5356,
+      "step": 11679
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513058721565372e-05,
+      "loss": 0.4633,
+      "step": 11680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512009416585601e-05,
+      "loss": 0.4807,
+      "step": 11681
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3510960067494794e-05,
+      "loss": 0.4519,
+      "step": 11682
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509910674306133e-05,
+      "loss": 0.4447,
+      "step": 11683
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3508861237032796e-05,
+      "loss": 0.4655,
+      "step": 11684
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3507811755687965e-05,
+      "loss": 0.5306,
+      "step": 11685
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506762230284823e-05,
+      "loss": 0.469,
+      "step": 11686
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350571266083655e-05,
+      "loss": 0.4778,
+      "step": 11687
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504663047356331e-05,
+      "loss": 0.4773,
+      "step": 11688
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503613389857349e-05,
+      "loss": 0.5002,
+      "step": 11689
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502563688352784e-05,
+      "loss": 0.5183,
+      "step": 11690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501513942855821e-05,
+      "loss": 0.4393,
+      "step": 11691
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350046415337965e-05,
+      "loss": 0.4786,
+      "step": 11692
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499414319937456e-05,
+      "loss": 0.4747,
+      "step": 11693
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498364442542417e-05,
+      "loss": 0.4583,
+      "step": 11694
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3497314521207728e-05,
+      "loss": 0.4523,
+      "step": 11695
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496264555946571e-05,
+      "loss": 0.4841,
+      "step": 11696
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3495214546772135e-05,
+      "loss": 0.5263,
+      "step": 11697
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3494164493697612e-05,
+      "loss": 0.4665,
+      "step": 11698
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3493114396736181e-05,
+      "loss": 0.509,
+      "step": 11699
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492064255901043e-05,
+      "loss": 0.4162,
+      "step": 11700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349101407120538e-05,
+      "loss": 0.4631,
+      "step": 11701
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489963842662384e-05,
+      "loss": 0.4462,
+      "step": 11702
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348891357028525e-05,
+      "loss": 0.4732,
+      "step": 11703
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3487863254087164e-05,
+      "loss": 0.4437,
+      "step": 11704
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486812894081318e-05,
+      "loss": 0.467,
+      "step": 11705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3485762490280911e-05,
+      "loss": 0.4804,
+      "step": 11706
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484712042699128e-05,
+      "loss": 0.4705,
+      "step": 11707
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3483661551349169e-05,
+      "loss": 0.5131,
+      "step": 11708
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3482611016244227e-05,
+      "loss": 0.416,
+      "step": 11709
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481560437397494e-05,
+      "loss": 0.4438,
+      "step": 11710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480509814822165e-05,
+      "loss": 0.4961,
+      "step": 11711
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347945914853144e-05,
+      "loss": 0.4567,
+      "step": 11712
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478408438538513e-05,
+      "loss": 0.4502,
+      "step": 11713
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347735768485658e-05,
+      "loss": 0.5247,
+      "step": 11714
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3476306887498842e-05,
+      "loss": 0.4083,
+      "step": 11715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475256046478493e-05,
+      "loss": 0.4664,
+      "step": 11716
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3474205161808734e-05,
+      "loss": 0.4674,
+      "step": 11717
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473154233502763e-05,
+      "loss": 0.4461,
+      "step": 11718
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472103261573782e-05,
+      "loss": 0.477,
+      "step": 11719
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3471052246034988e-05,
+      "loss": 0.4804,
+      "step": 11720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470001186899584e-05,
+      "loss": 0.5043,
+      "step": 11721
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346895008418077e-05,
+      "loss": 0.3975,
+      "step": 11722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3467898937891749e-05,
+      "loss": 0.4509,
+      "step": 11723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466847748045726e-05,
+      "loss": 0.4844,
+      "step": 11724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3465796514655895e-05,
+      "loss": 0.4498,
+      "step": 11725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346474523773547e-05,
+      "loss": 0.4635,
+      "step": 11726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463693917297651e-05,
+      "loss": 0.4469,
+      "step": 11727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346264255335564e-05,
+      "loss": 0.4585,
+      "step": 11728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3461591145922647e-05,
+      "loss": 0.4545,
+      "step": 11729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460539695011874e-05,
+      "loss": 0.455,
+      "step": 11730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459488200636527e-05,
+      "loss": 0.4682,
+      "step": 11731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3458436662809817e-05,
+      "loss": 0.4537,
+      "step": 11732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345738508154495e-05,
+      "loss": 0.4153,
+      "step": 11733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345633345685513e-05,
+      "loss": 0.4558,
+      "step": 11734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345528178875357e-05,
+      "loss": 0.4562,
+      "step": 11735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454230077253473e-05,
+      "loss": 0.4584,
+      "step": 11736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345317832236806e-05,
+      "loss": 0.4502,
+      "step": 11737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452126524110529e-05,
+      "loss": 0.5218,
+      "step": 11738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451074682494096e-05,
+      "loss": 0.4738,
+      "step": 11739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450022797531973e-05,
+      "loss": 0.447,
+      "step": 11740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344897086923737e-05,
+      "loss": 0.4585,
+      "step": 11741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447918897623497e-05,
+      "loss": 0.4442,
+      "step": 11742
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3446866882703572e-05,
+      "loss": 0.5285,
+      "step": 11743
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3445814824490806e-05,
+      "loss": 0.4774,
+      "step": 11744
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444762722998413e-05,
+      "loss": 0.4573,
+      "step": 11745
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443710578239607e-05,
+      "loss": 0.4759,
+      "step": 11746
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442658390227604e-05,
+      "loss": 0.4957,
+      "step": 11747
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3441606158975616e-05,
+      "loss": 0.4678,
+      "step": 11748
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440553884496863e-05,
+      "loss": 0.4846,
+      "step": 11749
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439501566804563e-05,
+      "loss": 0.4605,
+      "step": 11750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343844920591193e-05,
+      "loss": 0.4463,
+      "step": 11751
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437396801832181e-05,
+      "loss": 0.4642,
+      "step": 11752
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436344354578536e-05,
+      "loss": 0.4307,
+      "step": 11753
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435291864164213e-05,
+      "loss": 0.4271,
+      "step": 11754
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3434239330602432e-05,
+      "loss": 0.4606,
+      "step": 11755
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3433186753906413e-05,
+      "loss": 0.5042,
+      "step": 11756
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3432134134089376e-05,
+      "loss": 0.4686,
+      "step": 11757
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431081471164542e-05,
+      "loss": 0.4727,
+      "step": 11758
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343002876514513e-05,
+      "loss": 0.4805,
+      "step": 11759
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3428976016044368e-05,
+      "loss": 0.4891,
+      "step": 11760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427923223875479e-05,
+      "loss": 0.5115,
+      "step": 11761
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3426870388651676e-05,
+      "loss": 0.471,
+      "step": 11762
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342581751038619e-05,
+      "loss": 0.4993,
+      "step": 11763
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3424764589092245e-05,
+      "loss": 0.5,
+      "step": 11764
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3423711624783064e-05,
+      "loss": 0.4229,
+      "step": 11765
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422658617471878e-05,
+      "loss": 0.4823,
+      "step": 11766
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3421605567171902e-05,
+      "loss": 0.4274,
+      "step": 11767
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342055247389637e-05,
+      "loss": 0.4826,
+      "step": 11768
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419499337658508e-05,
+      "loss": 0.4455,
+      "step": 11769
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3418446158471542e-05,
+      "loss": 0.4128,
+      "step": 11770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417392936348703e-05,
+      "loss": 0.4851,
+      "step": 11771
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3416339671303215e-05,
+      "loss": 0.4561,
+      "step": 11772
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341528636334831e-05,
+      "loss": 0.4082,
+      "step": 11773
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341423301249722e-05,
+      "loss": 0.4307,
+      "step": 11774
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413179618763168e-05,
+      "loss": 0.4739,
+      "step": 11775
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412126182159391e-05,
+      "loss": 0.5268,
+      "step": 11776
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411072702699117e-05,
+      "loss": 0.4648,
+      "step": 11777
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410019180395582e-05,
+      "loss": 0.464,
+      "step": 11778
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3408965615262008e-05,
+      "loss": 0.4771,
+      "step": 11779
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407912007311641e-05,
+      "loss": 0.4706,
+      "step": 11780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406858356557706e-05,
+      "loss": 0.4427,
+      "step": 11781
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405804663013439e-05,
+      "loss": 0.4714,
+      "step": 11782
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3404750926692077e-05,
+      "loss": 0.4468,
+      "step": 11783
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3403697147606853e-05,
+      "loss": 0.4578,
+      "step": 11784
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402643325770996e-05,
+      "loss": 0.4956,
+      "step": 11785
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401589461197755e-05,
+      "loss": 0.4654,
+      "step": 11786
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400535553900357e-05,
+      "loss": 0.4596,
+      "step": 11787
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3399481603892044e-05,
+      "loss": 0.4644,
+      "step": 11788
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398427611186051e-05,
+      "loss": 0.5045,
+      "step": 11789
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397373575795614e-05,
+      "loss": 0.4936,
+      "step": 11790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3396319497733976e-05,
+      "loss": 0.4612,
+      "step": 11791
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3395265377014375e-05,
+      "loss": 0.4385,
+      "step": 11792
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394211213650052e-05,
+      "loss": 0.4646,
+      "step": 11793
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3393157007654245e-05,
+      "loss": 0.4459,
+      "step": 11794
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392102759040197e-05,
+      "loss": 0.5293,
+      "step": 11795
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391048467821146e-05,
+      "loss": 0.4682,
+      "step": 11796
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389994134010339e-05,
+      "loss": 0.4588,
+      "step": 11797
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3388939757621018e-05,
+      "loss": 0.4318,
+      "step": 11798
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3387885338666417e-05,
+      "loss": 0.4856,
+      "step": 11799
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338683087715979e-05,
+      "loss": 0.5239,
+      "step": 11800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3385776373114379e-05,
+      "loss": 0.5178,
+      "step": 11801
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3384721826543423e-05,
+      "loss": 0.5216,
+      "step": 11802
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3383667237460178e-05,
+      "loss": 0.4449,
+      "step": 11803
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382612605877879e-05,
+      "loss": 0.4327,
+      "step": 11804
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381557931809774e-05,
+      "loss": 0.4241,
+      "step": 11805
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3380503215269114e-05,
+      "loss": 0.4405,
+      "step": 11806
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3379448456269146e-05,
+      "loss": 0.4955,
+      "step": 11807
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378393654823114e-05,
+      "loss": 0.4136,
+      "step": 11808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377338810944269e-05,
+      "loss": 0.4913,
+      "step": 11809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3376283924645859e-05,
+      "loss": 0.4668,
+      "step": 11810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375228995941135e-05,
+      "loss": 0.485,
+      "step": 11811
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374174024843345e-05,
+      "loss": 0.4401,
+      "step": 11812
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337311901136574e-05,
+      "loss": 0.4817,
+      "step": 11813
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372063955521573e-05,
+      "loss": 0.4419,
+      "step": 11814
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371008857324092e-05,
+      "loss": 0.463,
+      "step": 11815
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336995371678655e-05,
+      "loss": 0.4673,
+      "step": 11816
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368898533922202e-05,
+      "loss": 0.4283,
+      "step": 11817
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367843308744299e-05,
+      "loss": 0.4664,
+      "step": 11818
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366788041266096e-05,
+      "loss": 0.4944,
+      "step": 11819
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3365732731500849e-05,
+      "loss": 0.5027,
+      "step": 11820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364677379461809e-05,
+      "loss": 0.4538,
+      "step": 11821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336362198516223e-05,
+      "loss": 0.4827,
+      "step": 11822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3362566548615373e-05,
+      "loss": 0.487,
+      "step": 11823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336151106983449e-05,
+      "loss": 0.4428,
+      "step": 11824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336045554883284e-05,
+      "loss": 0.434,
+      "step": 11825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359399985623682e-05,
+      "loss": 0.4915,
+      "step": 11826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335834438022027e-05,
+      "loss": 0.4083,
+      "step": 11827
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357288732635864e-05,
+      "loss": 0.4746,
+      "step": 11828
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356233042883726e-05,
+      "loss": 0.4638,
+      "step": 11829
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355177310977112e-05,
+      "loss": 0.5233,
+      "step": 11830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354121536929285e-05,
+      "loss": 0.4989,
+      "step": 11831
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3353065720753501e-05,
+      "loss": 0.4318,
+      "step": 11832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352009862463023e-05,
+      "loss": 0.4534,
+      "step": 11833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350953962071115e-05,
+      "loss": 0.4358,
+      "step": 11834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349898019591039e-05,
+      "loss": 0.4501,
+      "step": 11835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348842035036054e-05,
+      "loss": 0.4992,
+      "step": 11836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347786008419426e-05,
+      "loss": 0.5136,
+      "step": 11837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334672993975442e-05,
+      "loss": 0.499,
+      "step": 11838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345673829054295e-05,
+      "loss": 0.4836,
+      "step": 11839
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344617676332323e-05,
+      "loss": 0.4708,
+      "step": 11840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3343561481601763e-05,
+      "loss": 0.5074,
+      "step": 11841
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342505244875883e-05,
+      "loss": 0.4838,
+      "step": 11842
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341448966167954e-05,
+      "loss": 0.4807,
+      "step": 11843
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340392645491234e-05,
+      "loss": 0.4356,
+      "step": 11844
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3339336282858999e-05,
+      "loss": 0.4047,
+      "step": 11845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338279878284512e-05,
+      "loss": 0.4937,
+      "step": 11846
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337223431781041e-05,
+      "loss": 0.5291,
+      "step": 11847
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336166943361857e-05,
+      "loss": 0.4445,
+      "step": 11848
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333511041304023e-05,
+      "loss": 0.4371,
+      "step": 11849
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334053840829434e-05,
+      "loss": 0.4867,
+      "step": 11850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333299722674273e-05,
+      "loss": 0.4895,
+      "step": 11851
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3331940570793395e-05,
+      "loss": 0.5211,
+      "step": 11852
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33308838729947e-05,
+      "loss": 0.4826,
+      "step": 11853
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329827133359916e-05,
+      "loss": 0.4559,
+      "step": 11854
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3328770351902318e-05,
+      "loss": 0.5064,
+      "step": 11855
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327713528635178e-05,
+      "loss": 0.5011,
+      "step": 11856
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332665666357177e-05,
+      "loss": 0.5056,
+      "step": 11857
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332559975672537e-05,
+      "loss": 0.4538,
+      "step": 11858
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324542808109245e-05,
+      "loss": 0.496,
+      "step": 11859
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323485817736679e-05,
+      "loss": 0.4961,
+      "step": 11860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322428785620947e-05,
+      "loss": 0.478,
+      "step": 11861
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321371711775321e-05,
+      "loss": 0.4396,
+      "step": 11862
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320314596213081e-05,
+      "loss": 0.4327,
+      "step": 11863
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3319257438947502e-05,
+      "loss": 0.508,
+      "step": 11864
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318200239991865e-05,
+      "loss": 0.4053,
+      "step": 11865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317142999359451e-05,
+      "loss": 0.4205,
+      "step": 11866
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331608571706353e-05,
+      "loss": 0.5003,
+      "step": 11867
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315028393117388e-05,
+      "loss": 0.4598,
+      "step": 11868
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3313971027534304e-05,
+      "loss": 0.4669,
+      "step": 11869
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312913620327557e-05,
+      "loss": 0.4421,
+      "step": 11870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311856171510433e-05,
+      "loss": 0.4003,
+      "step": 11871
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3310798681096209e-05,
+      "loss": 0.4464,
+      "step": 11872
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309741149098166e-05,
+      "loss": 0.4616,
+      "step": 11873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3308683575529589e-05,
+      "loss": 0.5091,
+      "step": 11874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307625960403763e-05,
+      "loss": 0.4408,
+      "step": 11875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3306568303733966e-05,
+      "loss": 0.4831,
+      "step": 11876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3305510605533489e-05,
+      "loss": 0.4865,
+      "step": 11877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304452865815613e-05,
+      "loss": 0.4472,
+      "step": 11878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303395084593624e-05,
+      "loss": 0.4858,
+      "step": 11879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302337261880806e-05,
+      "loss": 0.4505,
+      "step": 11880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330127939769045e-05,
+      "loss": 0.4935,
+      "step": 11881
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300221492035838e-05,
+      "loss": 0.4537,
+      "step": 11882
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299163544930263e-05,
+      "loss": 0.4214,
+      "step": 11883
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298105556387003e-05,
+      "loss": 0.422,
+      "step": 11884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3297047526419356e-05,
+      "loss": 0.4468,
+      "step": 11885
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3295989455040608e-05,
+      "loss": 0.4119,
+      "step": 11886
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3294931342264046e-05,
+      "loss": 0.4482,
+      "step": 11887
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329387318810296e-05,
+      "loss": 0.4973,
+      "step": 11888
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292814992570645e-05,
+      "loss": 0.4325,
+      "step": 11889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329175675568039e-05,
+      "loss": 0.4468,
+      "step": 11890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290698477445485e-05,
+      "loss": 0.4939,
+      "step": 11891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289640157879222e-05,
+      "loss": 0.4423,
+      "step": 11892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3288581796994894e-05,
+      "loss": 0.4367,
+      "step": 11893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.32875233948058e-05,
+      "loss": 0.4571,
+      "step": 11894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3286464951325223e-05,
+      "loss": 0.4911,
+      "step": 11895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328540646656646e-05,
+      "loss": 0.4458,
+      "step": 11896
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328434794054281e-05,
+      "loss": 0.4523,
+      "step": 11897
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283289373267569e-05,
+      "loss": 0.3668,
+      "step": 11898
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3282230764754025e-05,
+      "loss": 0.4408,
+      "step": 11899
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281172115015481e-05,
+      "loss": 0.439,
+      "step": 11900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3280113424065228e-05,
+      "loss": 0.451,
+      "step": 11901
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279054691916572e-05,
+      "loss": 0.4736,
+      "step": 11902
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3277995918582805e-05,
+      "loss": 0.3868,
+      "step": 11903
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276937104077223e-05,
+      "loss": 0.4708,
+      "step": 11904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275878248413128e-05,
+      "loss": 0.4853,
+      "step": 11905
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274819351603821e-05,
+      "loss": 0.4157,
+      "step": 11906
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273760413662596e-05,
+      "loss": 0.5029,
+      "step": 11907
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327270143460276e-05,
+      "loss": 0.4782,
+      "step": 11908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3271642414437609e-05,
+      "loss": 0.4521,
+      "step": 11909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270583353180445e-05,
+      "loss": 0.4499,
+      "step": 11910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269524250844572e-05,
+      "loss": 0.4567,
+      "step": 11911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3268465107443291e-05,
+      "loss": 0.4597,
+      "step": 11912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267405922989907e-05,
+      "loss": 0.43,
+      "step": 11913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266346697497721e-05,
+      "loss": 0.4509,
+      "step": 11914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265287430980036e-05,
+      "loss": 0.4643,
+      "step": 11915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264228123450159e-05,
+      "loss": 0.4316,
+      "step": 11916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3263168774921393e-05,
+      "loss": 0.4796,
+      "step": 11917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262109385407044e-05,
+      "loss": 0.4444,
+      "step": 11918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261049954920423e-05,
+      "loss": 0.4639,
+      "step": 11919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3259990483474827e-05,
+      "loss": 0.4566,
+      "step": 11920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258930971083568e-05,
+      "loss": 0.434,
+      "step": 11921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257871417759956e-05,
+      "loss": 0.4623,
+      "step": 11922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256811823517295e-05,
+      "loss": 0.4526,
+      "step": 11923
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3255752188368896e-05,
+      "loss": 0.4408,
+      "step": 11924
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254692512328066e-05,
+      "loss": 0.4778,
+      "step": 11925
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253632795408118e-05,
+      "loss": 0.4468,
+      "step": 11926
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252573037622356e-05,
+      "loss": 0.4751,
+      "step": 11927
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251513238984097e-05,
+      "loss": 0.5223,
+      "step": 11928
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250453399506648e-05,
+      "loss": 0.4896,
+      "step": 11929
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249393519203326e-05,
+      "loss": 0.4704,
+      "step": 11930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248333598087437e-05,
+      "loss": 0.5355,
+      "step": 11931
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247273636172295e-05,
+      "loss": 0.4484,
+      "step": 11932
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3246213633471214e-05,
+      "loss": 0.4735,
+      "step": 11933
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324515358999751e-05,
+      "loss": 0.4373,
+      "step": 11934
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32440935057645e-05,
+      "loss": 0.4783,
+      "step": 11935
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243033380785485e-05,
+      "loss": 0.4506,
+      "step": 11936
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241973215073794e-05,
+      "loss": 0.4598,
+      "step": 11937
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240913008642732e-05,
+      "loss": 0.4834,
+      "step": 11938
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3239852761505627e-05,
+      "loss": 0.4687,
+      "step": 11939
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238792473675792e-05,
+      "loss": 0.4404,
+      "step": 11940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237732145166536e-05,
+      "loss": 0.4726,
+      "step": 11941
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236671775991185e-05,
+      "loss": 0.4615,
+      "step": 11942
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3235611366163057e-05,
+      "loss": 0.4495,
+      "step": 11943
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3234550915695466e-05,
+      "loss": 0.4844,
+      "step": 11944
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233490424601739e-05,
+      "loss": 0.4569,
+      "step": 11945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232429892895187e-05,
+      "loss": 0.4809,
+      "step": 11946
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231369320589133e-05,
+      "loss": 0.492,
+      "step": 11947
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230308707696905e-05,
+      "loss": 0.4596,
+      "step": 11948
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3229248054231816e-05,
+      "loss": 0.5107,
+      "step": 11949
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228187360207189e-05,
+      "loss": 0.4724,
+      "step": 11950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322712662563635e-05,
+      "loss": 0.4903,
+      "step": 11951
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322606585053262e-05,
+      "loss": 0.4751,
+      "step": 11952
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322500503490932e-05,
+      "loss": 0.5089,
+      "step": 11953
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223944178779777e-05,
+      "loss": 0.4893,
+      "step": 11954
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222883282157315e-05,
+      "loss": 0.4792,
+      "step": 11955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221822345055259e-05,
+      "loss": 0.4622,
+      "step": 11956
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220761367486934e-05,
+      "loss": 0.4268,
+      "step": 11957
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219700349465665e-05,
+      "loss": 0.492,
+      "step": 11958
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321863929100478e-05,
+      "loss": 0.4613,
+      "step": 11959
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3217578192117605e-05,
+      "loss": 0.4724,
+      "step": 11960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216517052817466e-05,
+      "loss": 0.4525,
+      "step": 11961
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215455873117695e-05,
+      "loss": 0.4502,
+      "step": 11962
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.4585,
+      "step": 11963
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213333392572562e-05,
+      "loss": 0.4842,
+      "step": 11964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3212272091753859e-05,
+      "loss": 0.4524,
+      "step": 11965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211210750588841e-05,
+      "loss": 0.4001,
+      "step": 11966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210149369090837e-05,
+      "loss": 0.491,
+      "step": 11967
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209087947273174e-05,
+      "loss": 0.466,
+      "step": 11968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3208026485149189e-05,
+      "loss": 0.4606,
+      "step": 11969
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206964982732207e-05,
+      "loss": 0.4631,
+      "step": 11970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320590344003557e-05,
+      "loss": 0.4942,
+      "step": 11971
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204841857072608e-05,
+      "loss": 0.4454,
+      "step": 11972
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203780233856646e-05,
+      "loss": 0.5123,
+      "step": 11973
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320271857040103e-05,
+      "loss": 0.5103,
+      "step": 11974
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201656866719086e-05,
+      "loss": 0.4935,
+      "step": 11975
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3200595122824153e-05,
+      "loss": 0.4566,
+      "step": 11976
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199533338729568e-05,
+      "loss": 0.5024,
+      "step": 11977
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198471514448665e-05,
+      "loss": 0.4863,
+      "step": 11978
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319740964999478e-05,
+      "loss": 0.5527,
+      "step": 11979
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3196347745381251e-05,
+      "loss": 0.483,
+      "step": 11980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195285800621413e-05,
+      "loss": 0.4922,
+      "step": 11981
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3194223815728611e-05,
+      "loss": 0.491,
+      "step": 11982
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193161790716176e-05,
+      "loss": 0.4682,
+      "step": 11983
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319209972559745e-05,
+      "loss": 0.4638,
+      "step": 11984
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191037620385774e-05,
+      "loss": 0.4447,
+      "step": 11985
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189975475094486e-05,
+      "loss": 0.5058,
+      "step": 11986
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188913289736926e-05,
+      "loss": 0.4626,
+      "step": 11987
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187851064326437e-05,
+      "loss": 0.4468,
+      "step": 11988
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186788798876362e-05,
+      "loss": 0.485,
+      "step": 11989
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185726493400037e-05,
+      "loss": 0.4665,
+      "step": 11990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3184664147910815e-05,
+      "loss": 0.4845,
+      "step": 11991
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.318360176242203e-05,
+      "loss": 0.4407,
+      "step": 11992
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3182539336947026e-05,
+      "loss": 0.4349,
+      "step": 11993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3181476871499155e-05,
+      "loss": 0.4783,
+      "step": 11994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3180414366091751e-05,
+      "loss": 0.4547,
+      "step": 11995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179351820738167e-05,
+      "loss": 0.5146,
+      "step": 11996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178289235451746e-05,
+      "loss": 0.4809,
+      "step": 11997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177226610245835e-05,
+      "loss": 0.5049,
+      "step": 11998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176163945133777e-05,
+      "loss": 0.4767,
+      "step": 11999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3175101240128924e-05,
+      "loss": 0.5164,
+      "step": 12000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174038495244621e-05,
+      "loss": 0.5031,
+      "step": 12001
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3172975710494218e-05,
+      "loss": 0.4504,
+      "step": 12002
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171912885891063e-05,
+      "loss": 0.4012,
+      "step": 12003
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3170850021448502e-05,
+      "loss": 0.4413,
+      "step": 12004
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169787117179888e-05,
+      "loss": 0.5,
+      "step": 12005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168724173098571e-05,
+      "loss": 0.3957,
+      "step": 12006
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167661189217897e-05,
+      "loss": 0.4797,
+      "step": 12007
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166598165551227e-05,
+      "loss": 0.4182,
+      "step": 12008
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165535102111904e-05,
+      "loss": 0.4573,
+      "step": 12009
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3164471998913281e-05,
+      "loss": 0.4584,
+      "step": 12010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3163408855968718e-05,
+      "loss": 0.4669,
+      "step": 12011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162345673291557e-05,
+      "loss": 0.4683,
+      "step": 12012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3161282450895162e-05,
+      "loss": 0.4784,
+      "step": 12013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316021918879288e-05,
+      "loss": 0.439,
+      "step": 12014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159155886998065e-05,
+      "loss": 0.4215,
+      "step": 12015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158092545524077e-05,
+      "loss": 0.3969,
+      "step": 12016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3157029164384272e-05,
+      "loss": 0.4519,
+      "step": 12017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155965743592001e-05,
+      "loss": 0.5014,
+      "step": 12018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3154902283160626e-05,
+      "loss": 0.4572,
+      "step": 12019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153838783103501e-05,
+      "loss": 0.5062,
+      "step": 12020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3152775243433981e-05,
+      "loss": 0.4262,
+      "step": 12021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3151711664165432e-05,
+      "loss": 0.4636,
+      "step": 12022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150648045311205e-05,
+      "loss": 0.4642,
+      "step": 12023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149584386884663e-05,
+      "loss": 0.4635,
+      "step": 12024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148520688899166e-05,
+      "loss": 0.4887,
+      "step": 12025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3147456951368072e-05,
+      "loss": 0.459,
+      "step": 12026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146393174304741e-05,
+      "loss": 0.5018,
+      "step": 12027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145329357722536e-05,
+      "loss": 0.4253,
+      "step": 12028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314426550163482e-05,
+      "loss": 0.4687,
+      "step": 12029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143201606054951e-05,
+      "loss": 0.4961,
+      "step": 12030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142137670996296e-05,
+      "loss": 0.4775,
+      "step": 12031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141073696472214e-05,
+      "loss": 0.4472,
+      "step": 12032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140009682496071e-05,
+      "loss": 0.4359,
+      "step": 12033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313894562908123e-05,
+      "loss": 0.4287,
+      "step": 12034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137881536241058e-05,
+      "loss": 0.4644,
+      "step": 12035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136817403988918e-05,
+      "loss": 0.5126,
+      "step": 12036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135753232338173e-05,
+      "loss": 0.4967,
+      "step": 12037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3134689021302192e-05,
+      "loss": 0.4747,
+      "step": 12038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133624770894345e-05,
+      "loss": 0.4783,
+      "step": 12039
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3132560481127995e-05,
+      "loss": 0.4706,
+      "step": 12040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3131496152016505e-05,
+      "loss": 0.4319,
+      "step": 12041
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313043178357325e-05,
+      "loss": 0.5202,
+      "step": 12042
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31293673758116e-05,
+      "loss": 0.4284,
+      "step": 12043
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128302928744914e-05,
+      "loss": 0.4854,
+      "step": 12044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3127238442386575e-05,
+      "loss": 0.4551,
+      "step": 12045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3126173916749943e-05,
+      "loss": 0.4224,
+      "step": 12046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125109351848391e-05,
+      "loss": 0.467,
+      "step": 12047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312404474769529e-05,
+      "loss": 0.4229,
+      "step": 12048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122980104304013e-05,
+      "loss": 0.4604,
+      "step": 12049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121915421687933e-05,
+      "loss": 0.4869,
+      "step": 12050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3120850699860417e-05,
+      "loss": 0.4511,
+      "step": 12051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3119785938834842e-05,
+      "loss": 0.4385,
+      "step": 12052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3118721138624581e-05,
+      "loss": 0.4704,
+      "step": 12053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311765629924301e-05,
+      "loss": 0.5028,
+      "step": 12054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116591420703499e-05,
+      "loss": 0.3824,
+      "step": 12055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115526503019426e-05,
+      "loss": 0.4578,
+      "step": 12056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3114461546204168e-05,
+      "loss": 0.424,
+      "step": 12057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113396550271093e-05,
+      "loss": 0.4383,
+      "step": 12058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3112331515233587e-05,
+      "loss": 0.449,
+      "step": 12059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111266441105021e-05,
+      "loss": 0.4337,
+      "step": 12060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3110201327898778e-05,
+      "loss": 0.5097,
+      "step": 12061
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109136175628229e-05,
+      "loss": 0.4809,
+      "step": 12062
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108070984306756e-05,
+      "loss": 0.4389,
+      "step": 12063
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107005753947735e-05,
+      "loss": 0.4752,
+      "step": 12064
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310594048456455e-05,
+      "loss": 0.4774,
+      "step": 12065
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310487517617058e-05,
+      "loss": 0.4592,
+      "step": 12066
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103809828779202e-05,
+      "loss": 0.4506,
+      "step": 12067
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102744442403798e-05,
+      "loss": 0.4456,
+      "step": 12068
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310167901705775e-05,
+      "loss": 0.4638,
+      "step": 12069
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100613552754442e-05,
+      "loss": 0.4345,
+      "step": 12070
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099548049507253e-05,
+      "loss": 0.4847,
+      "step": 12071
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098482507329568e-05,
+      "loss": 0.4313,
+      "step": 12072
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309741692623477e-05,
+      "loss": 0.5668,
+      "step": 12073
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096351306236239e-05,
+      "loss": 0.4533,
+      "step": 12074
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095285647347361e-05,
+      "loss": 0.4509,
+      "step": 12075
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3094219949581526e-05,
+      "loss": 0.4817,
+      "step": 12076
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093154212952116e-05,
+      "loss": 0.4523,
+      "step": 12077
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309208843747251e-05,
+      "loss": 0.4625,
+      "step": 12078
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3091022623156104e-05,
+      "loss": 0.492,
+      "step": 12079
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3089956770016281e-05,
+      "loss": 0.4661,
+      "step": 12080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088890878066425e-05,
+      "loss": 0.4305,
+      "step": 12081
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087824947319931e-05,
+      "loss": 0.4499,
+      "step": 12082
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3086758977790179e-05,
+      "loss": 0.5497,
+      "step": 12083
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085692969490563e-05,
+      "loss": 0.4647,
+      "step": 12084
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308462692243447e-05,
+      "loss": 0.4813,
+      "step": 12085
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083560836635288e-05,
+      "loss": 0.4633,
+      "step": 12086
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082494712106414e-05,
+      "loss": 0.4346,
+      "step": 12087
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308142854886123e-05,
+      "loss": 0.5038,
+      "step": 12088
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080362346913132e-05,
+      "loss": 0.4504,
+      "step": 12089
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307929610627551e-05,
+      "loss": 0.3623,
+      "step": 12090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3078229826961757e-05,
+      "loss": 0.4885,
+      "step": 12091
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3077163508985264e-05,
+      "loss": 0.4793,
+      "step": 12092
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076097152359428e-05,
+      "loss": 0.4065,
+      "step": 12093
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075030757097638e-05,
+      "loss": 0.4553,
+      "step": 12094
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073964323213285e-05,
+      "loss": 0.4894,
+      "step": 12095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072897850719772e-05,
+      "loss": 0.5235,
+      "step": 12096
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307183133963049e-05,
+      "loss": 0.4242,
+      "step": 12097
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070764789958834e-05,
+      "loss": 0.4806,
+      "step": 12098
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069698201718202e-05,
+      "loss": 0.4885,
+      "step": 12099
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3068631574921984e-05,
+      "loss": 0.5276,
+      "step": 12100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067564909583586e-05,
+      "loss": 0.4338,
+      "step": 12101
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066498205716399e-05,
+      "loss": 0.4562,
+      "step": 12102
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3065431463333824e-05,
+      "loss": 0.4848,
+      "step": 12103
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064364682449258e-05,
+      "loss": 0.4426,
+      "step": 12104
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30632978630761e-05,
+      "loss": 0.4444,
+      "step": 12105
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062231005227745e-05,
+      "loss": 0.4814,
+      "step": 12106
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3061164108917601e-05,
+      "loss": 0.5342,
+      "step": 12107
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3060097174159065e-05,
+      "loss": 0.4389,
+      "step": 12108
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059030200965536e-05,
+      "loss": 0.4971,
+      "step": 12109
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057963189350415e-05,
+      "loss": 0.484,
+      "step": 12110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3056896139327109e-05,
+      "loss": 0.4283,
+      "step": 12111
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055829050909012e-05,
+      "loss": 0.4934,
+      "step": 12112
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3054761924109531e-05,
+      "loss": 0.4523,
+      "step": 12113
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053694758942075e-05,
+      "loss": 0.431,
+      "step": 12114
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3052627555420037e-05,
+      "loss": 0.4599,
+      "step": 12115
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3051560313556826e-05,
+      "loss": 0.473,
+      "step": 12116
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3050493033365848e-05,
+      "loss": 0.4992,
+      "step": 12117
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049425714860502e-05,
+      "loss": 0.4644,
+      "step": 12118
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048358358054205e-05,
+      "loss": 0.4625,
+      "step": 12119
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047290962960352e-05,
+      "loss": 0.4796,
+      "step": 12120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046223529592355e-05,
+      "loss": 0.5178,
+      "step": 12121
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045156057963621e-05,
+      "loss": 0.4581,
+      "step": 12122
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3044088548087554e-05,
+      "loss": 0.4056,
+      "step": 12123
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043020999977567e-05,
+      "loss": 0.4515,
+      "step": 12124
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041953413647066e-05,
+      "loss": 0.4288,
+      "step": 12125
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040885789109456e-05,
+      "loss": 0.4539,
+      "step": 12126
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039818126378152e-05,
+      "loss": 0.4736,
+      "step": 12127
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3038750425466565e-05,
+      "loss": 0.4349,
+      "step": 12128
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30376826863881e-05,
+      "loss": 0.467,
+      "step": 12129
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3036614909156171e-05,
+      "loss": 0.4886,
+      "step": 12130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035547093784187e-05,
+      "loss": 0.4506,
+      "step": 12131
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034479240285562e-05,
+      "loss": 0.4559,
+      "step": 12132
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033411348673708e-05,
+      "loss": 0.4525,
+      "step": 12133
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3032343418962037e-05,
+      "loss": 0.4391,
+      "step": 12134
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3031275451163964e-05,
+      "loss": 0.4223,
+      "step": 12135
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030207445292903e-05,
+      "loss": 0.4097,
+      "step": 12136
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029139401362263e-05,
+      "loss": 0.467,
+      "step": 12137
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028071319385464e-05,
+      "loss": 0.4312,
+      "step": 12138
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3027003199375922e-05,
+      "loss": 0.4735,
+      "step": 12139
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025935041347048e-05,
+      "loss": 0.454,
+      "step": 12140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302486684531226e-05,
+      "loss": 0.4976,
+      "step": 12141
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023798611284977e-05,
+      "loss": 0.5062,
+      "step": 12142
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302273033927861e-05,
+      "loss": 0.5236,
+      "step": 12143
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021662029306585e-05,
+      "loss": 0.4223,
+      "step": 12144
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020593681382318e-05,
+      "loss": 0.465,
+      "step": 12145
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019525295519219e-05,
+      "loss": 0.4494,
+      "step": 12146
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3018456871730716e-05,
+      "loss": 0.4531,
+      "step": 12147
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017388410030226e-05,
+      "loss": 0.4999,
+      "step": 12148
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016319910431166e-05,
+      "loss": 0.4634,
+      "step": 12149
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3015251372946965e-05,
+      "loss": 0.4869,
+      "step": 12150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014182797591033e-05,
+      "loss": 0.4746,
+      "step": 12151
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3013114184376793e-05,
+      "loss": 0.5081,
+      "step": 12152
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012045533317676e-05,
+      "loss": 0.4514,
+      "step": 12153
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3010976844427097e-05,
+      "loss": 0.3888,
+      "step": 12154
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300990811771848e-05,
+      "loss": 0.4519,
+      "step": 12155
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300883935320525e-05,
+      "loss": 0.4382,
+      "step": 12156
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007770550900824e-05,
+      "loss": 0.3899,
+      "step": 12157
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006701710818634e-05,
+      "loss": 0.4836,
+      "step": 12158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005632832972104e-05,
+      "loss": 0.4597,
+      "step": 12159
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3004563917374657e-05,
+      "loss": 0.4668,
+      "step": 12160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003494964039717e-05,
+      "loss": 0.479,
+      "step": 12161
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002425972980715e-05,
+      "loss": 0.4387,
+      "step": 12162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300135694421107e-05,
+      "loss": 0.4866,
+      "step": 12163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000287877744218e-05,
+      "loss": 0.442,
+      "step": 12164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299921877359358e-05,
+      "loss": 0.4612,
+      "step": 12165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998149631772587e-05,
+      "loss": 0.4862,
+      "step": 12166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997080452294669e-05,
+      "loss": 0.49,
+      "step": 12167
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996011235173249e-05,
+      "loss": 0.4231,
+      "step": 12168
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299494198042176e-05,
+      "loss": 0.4484,
+      "step": 12169
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2993872688053633e-05,
+      "loss": 0.4441,
+      "step": 12170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992803358082298e-05,
+      "loss": 0.4647,
+      "step": 12171
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991733990521187e-05,
+      "loss": 0.467,
+      "step": 12172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2990664585383727e-05,
+      "loss": 0.4659,
+      "step": 12173
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298959514268335e-05,
+      "loss": 0.4909,
+      "step": 12174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988525662433495e-05,
+      "loss": 0.4422,
+      "step": 12175
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298745614464759e-05,
+      "loss": 0.4873,
+      "step": 12176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986386589339067e-05,
+      "loss": 0.52,
+      "step": 12177
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298531699652136e-05,
+      "loss": 0.4842,
+      "step": 12178
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2984247366207907e-05,
+      "loss": 0.4928,
+      "step": 12179
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983177698412136e-05,
+      "loss": 0.4984,
+      "step": 12180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298210799314749e-05,
+      "loss": 0.5106,
+      "step": 12181
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981038250427402e-05,
+      "loss": 0.4169,
+      "step": 12182
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2979968470265303e-05,
+      "loss": 0.4977,
+      "step": 12183
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2978898652674632e-05,
+      "loss": 0.4719,
+      "step": 12184
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.297782879766883e-05,
+      "loss": 0.4802,
+      "step": 12185
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2976758905261329e-05,
+      "loss": 0.512,
+      "step": 12186
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2975688975465572e-05,
+      "loss": 0.5198,
+      "step": 12187
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2974619008294995e-05,
+      "loss": 0.4112,
+      "step": 12188
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973549003763035e-05,
+      "loss": 0.465,
+      "step": 12189
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972478961883132e-05,
+      "loss": 0.4674,
+      "step": 12190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297140888266873e-05,
+      "loss": 0.5142,
+      "step": 12191
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2970338766133263e-05,
+      "loss": 0.4385,
+      "step": 12192
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969268612290177e-05,
+      "loss": 0.4853,
+      "step": 12193
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296819842115291e-05,
+      "loss": 0.4433,
+      "step": 12194
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967128192734903e-05,
+      "loss": 0.4426,
+      "step": 12195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966057927049602e-05,
+      "loss": 0.4214,
+      "step": 12196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2964987624110448e-05,
+      "loss": 0.4312,
+      "step": 12197
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296391728393088e-05,
+      "loss": 0.417,
+      "step": 12198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2962846906524348e-05,
+      "loss": 0.4368,
+      "step": 12199
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296177649190429e-05,
+      "loss": 0.4772,
+      "step": 12200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960706040084156e-05,
+      "loss": 0.4692,
+      "step": 12201
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959635551077387e-05,
+      "loss": 0.4264,
+      "step": 12202
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2958565024897431e-05,
+      "loss": 0.4593,
+      "step": 12203
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295749446155773e-05,
+      "loss": 0.4191,
+      "step": 12204
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2956423861071739e-05,
+      "loss": 0.4335,
+      "step": 12205
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2955353223452891e-05,
+      "loss": 0.5039,
+      "step": 12206
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2954282548714645e-05,
+      "loss": 0.4031,
+      "step": 12207
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953211836870446e-05,
+      "loss": 0.4591,
+      "step": 12208
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295214108793374e-05,
+      "loss": 0.3929,
+      "step": 12209
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951070301917974e-05,
+      "loss": 0.5119,
+      "step": 12210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949999478836599e-05,
+      "loss": 0.4635,
+      "step": 12211
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948928618703067e-05,
+      "loss": 0.4939,
+      "step": 12212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294785772153083e-05,
+      "loss": 0.4964,
+      "step": 12213
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2946786787333333e-05,
+      "loss": 0.4864,
+      "step": 12214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945715816124027e-05,
+      "loss": 0.4826,
+      "step": 12215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944644807916366e-05,
+      "loss": 0.5367,
+      "step": 12216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29435737627238e-05,
+      "loss": 0.4766,
+      "step": 12217
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2942502680559786e-05,
+      "loss": 0.3475,
+      "step": 12218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2941431561437772e-05,
+      "loss": 0.4424,
+      "step": 12219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940360405371211e-05,
+      "loss": 0.4746,
+      "step": 12220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293928921237356e-05,
+      "loss": 0.4027,
+      "step": 12221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2938217982458275e-05,
+      "loss": 0.4899,
+      "step": 12222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937146715638802e-05,
+      "loss": 0.4749,
+      "step": 12223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936075411928605e-05,
+      "loss": 0.4735,
+      "step": 12224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293500407134114e-05,
+      "loss": 0.4269,
+      "step": 12225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933932693889852e-05,
+      "loss": 0.4563,
+      "step": 12226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293286127958821e-05,
+      "loss": 0.4617,
+      "step": 12227
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2931789828449661e-05,
+      "loss": 0.433,
+      "step": 12228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930718340487674e-05,
+      "loss": 0.4722,
+      "step": 12229
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2929646815715698e-05,
+      "loss": 0.4437,
+      "step": 12230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928575254147191e-05,
+      "loss": 0.4424,
+      "step": 12231
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927503655795616e-05,
+      "loss": 0.4471,
+      "step": 12232
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926432020674436e-05,
+      "loss": 0.4458,
+      "step": 12233
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925360348797101e-05,
+      "loss": 0.4652,
+      "step": 12234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2924288640177078e-05,
+      "loss": 0.4738,
+      "step": 12235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923216894827825e-05,
+      "loss": 0.4177,
+      "step": 12236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922145112762805e-05,
+      "loss": 0.4117,
+      "step": 12237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2921073293995478e-05,
+      "loss": 0.4541,
+      "step": 12238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920001438539307e-05,
+      "loss": 0.448,
+      "step": 12239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918929546407754e-05,
+      "loss": 0.4916,
+      "step": 12240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917857617614285e-05,
+      "loss": 0.449,
+      "step": 12241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916785652172359e-05,
+      "loss": 0.4926,
+      "step": 12242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915713650095439e-05,
+      "loss": 0.4682,
+      "step": 12243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914641611396994e-05,
+      "loss": 0.4499,
+      "step": 12244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291356953609049e-05,
+      "loss": 0.4754,
+      "step": 12245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912497424189388e-05,
+      "loss": 0.5253,
+      "step": 12246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2911425275707154e-05,
+      "loss": 0.446,
+      "step": 12247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910353090657254e-05,
+      "loss": 0.5095,
+      "step": 12248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290928086905316e-05,
+      "loss": 0.4445,
+      "step": 12249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908208610908332e-05,
+      "loss": 0.4927,
+      "step": 12250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907136316236245e-05,
+      "loss": 0.4629,
+      "step": 12251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906063985050361e-05,
+      "loss": 0.5144,
+      "step": 12252
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2904991617364151e-05,
+      "loss": 0.4253,
+      "step": 12253
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290391921319108e-05,
+      "loss": 0.4692,
+      "step": 12254
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.4602,
+      "step": 12255
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2901774295438253e-05,
+      "loss": 0.46,
+      "step": 12256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290070178188543e-05,
+      "loss": 0.4323,
+      "step": 12257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289962923189963e-05,
+      "loss": 0.4308,
+      "step": 12258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2898556645494327e-05,
+      "loss": 0.4427,
+      "step": 12259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289748402268299e-05,
+      "loss": 0.485,
+      "step": 12260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896411363479091e-05,
+      "loss": 0.4856,
+      "step": 12261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895338667896103e-05,
+      "loss": 0.4493,
+      "step": 12262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2894265935947497e-05,
+      "loss": 0.4325,
+      "step": 12263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289319316764675e-05,
+      "loss": 0.4948,
+      "step": 12264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2892120363007335e-05,
+      "loss": 0.4162,
+      "step": 12265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891047522042727e-05,
+      "loss": 0.4329,
+      "step": 12266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889974644766402e-05,
+      "loss": 0.4453,
+      "step": 12267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888901731191828e-05,
+      "loss": 0.4651,
+      "step": 12268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288782878133249e-05,
+      "loss": 0.476,
+      "step": 12269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288675579520186e-05,
+      "loss": 0.4614,
+      "step": 12270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2885682772813417e-05,
+      "loss": 0.4177,
+      "step": 12271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884609714180635e-05,
+      "loss": 0.4533,
+      "step": 12272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883536619316996e-05,
+      "loss": 0.4551,
+      "step": 12273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882463488235972e-05,
+      "loss": 0.4547,
+      "step": 12274
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2881390320951046e-05,
+      "loss": 0.423,
+      "step": 12275
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880317117475698e-05,
+      "loss": 0.4439,
+      "step": 12276
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2879243877823404e-05,
+      "loss": 0.47,
+      "step": 12277
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2878170602007647e-05,
+      "loss": 0.4639,
+      "step": 12278
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877097290041907e-05,
+      "loss": 0.4542,
+      "step": 12279
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.287602394193966e-05,
+      "loss": 0.4316,
+      "step": 12280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2874950557714395e-05,
+      "loss": 0.4677,
+      "step": 12281
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2873877137379589e-05,
+      "loss": 0.4507,
+      "step": 12282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2872803680948724e-05,
+      "loss": 0.4488,
+      "step": 12283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871730188435286e-05,
+      "loss": 0.4278,
+      "step": 12284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2870656659852752e-05,
+      "loss": 0.453,
+      "step": 12285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2869583095214612e-05,
+      "loss": 0.4508,
+      "step": 12286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868509494534352e-05,
+      "loss": 0.5237,
+      "step": 12287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867435857825445e-05,
+      "loss": 0.4761,
+      "step": 12288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866362185101387e-05,
+      "loss": 0.4894,
+      "step": 12289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2865288476375657e-05,
+      "loss": 0.4551,
+      "step": 12290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864214731661743e-05,
+      "loss": 0.4263,
+      "step": 12291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863140950973136e-05,
+      "loss": 0.4376,
+      "step": 12292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862067134323318e-05,
+      "loss": 0.523,
+      "step": 12293
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286099328172577e-05,
+      "loss": 0.4256,
+      "step": 12294
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285991939319399e-05,
+      "loss": 0.4535,
+      "step": 12295
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858845468741463e-05,
+      "loss": 0.4599,
+      "step": 12296
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2857771508381678e-05,
+      "loss": 0.4822,
+      "step": 12297
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856697512128121e-05,
+      "loss": 0.4373,
+      "step": 12298
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2855623479994283e-05,
+      "loss": 0.4776,
+      "step": 12299
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854549411993654e-05,
+      "loss": 0.4526,
+      "step": 12300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853475308139725e-05,
+      "loss": 0.446,
+      "step": 12301
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852401168445985e-05,
+      "loss": 0.4558,
+      "step": 12302
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851326992925927e-05,
+      "loss": 0.4788,
+      "step": 12303
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850252781593047e-05,
+      "loss": 0.423,
+      "step": 12304
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2849178534460824e-05,
+      "loss": 0.4912,
+      "step": 12305
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848104251542765e-05,
+      "loss": 0.4329,
+      "step": 12306
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2847029932852357e-05,
+      "loss": 0.4963,
+      "step": 12307
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2845955578403091e-05,
+      "loss": 0.437,
+      "step": 12308
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844881188208467e-05,
+      "loss": 0.4666,
+      "step": 12309
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2843806762281972e-05,
+      "loss": 0.4534,
+      "step": 12310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842732300637107e-05,
+      "loss": 0.4553,
+      "step": 12311
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841657803287363e-05,
+      "loss": 0.4597,
+      "step": 12312
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284058327024624e-05,
+      "loss": 0.4662,
+      "step": 12313
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839508701527232e-05,
+      "loss": 0.4286,
+      "step": 12314
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838434097143836e-05,
+      "loss": 0.4216,
+      "step": 12315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2837359457109548e-05,
+      "loss": 0.4311,
+      "step": 12316
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836284781437864e-05,
+      "loss": 0.4089,
+      "step": 12317
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2835210070142288e-05,
+      "loss": 0.5231,
+      "step": 12318
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834135323236315e-05,
+      "loss": 0.5029,
+      "step": 12319
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283306054073344e-05,
+      "loss": 0.4318,
+      "step": 12320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283198572264717e-05,
+      "loss": 0.4857,
+      "step": 12321
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830910868990996e-05,
+      "loss": 0.487,
+      "step": 12322
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2829835979778424e-05,
+      "loss": 0.4904,
+      "step": 12323
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828761055022959e-05,
+      "loss": 0.4818,
+      "step": 12324
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282768609473809e-05,
+      "loss": 0.5054,
+      "step": 12325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282661109893733e-05,
+      "loss": 0.445,
+      "step": 12326
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2825536067634174e-05,
+      "loss": 0.4291,
+      "step": 12327
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824461000842126e-05,
+      "loss": 0.4856,
+      "step": 12328
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2823385898574693e-05,
+      "loss": 0.4745,
+      "step": 12329
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822310760845373e-05,
+      "loss": 0.461,
+      "step": 12330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2821235587667668e-05,
+      "loss": 0.4436,
+      "step": 12331
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820160379055091e-05,
+      "loss": 0.5237,
+      "step": 12332
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819085135021141e-05,
+      "loss": 0.4798,
+      "step": 12333
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2818009855579323e-05,
+      "loss": 0.4943,
+      "step": 12334
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816934540743144e-05,
+      "loss": 0.4493,
+      "step": 12335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2815859190526107e-05,
+      "loss": 0.4404,
+      "step": 12336
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814783804941722e-05,
+      "loss": 0.439,
+      "step": 12337
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813708384003495e-05,
+      "loss": 0.4524,
+      "step": 12338
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812632927724935e-05,
+      "loss": 0.5091,
+      "step": 12339
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2811557436119548e-05,
+      "loss": 0.468,
+      "step": 12340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281048190920084e-05,
+      "loss": 0.4967,
+      "step": 12341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280940634698232e-05,
+      "loss": 0.5259,
+      "step": 12342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808330749477502e-05,
+      "loss": 0.4789,
+      "step": 12343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807255116699892e-05,
+      "loss": 0.4925,
+      "step": 12344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806179448663e-05,
+      "loss": 0.4464,
+      "step": 12345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280510374538034e-05,
+      "loss": 0.4251,
+      "step": 12346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804028006865417e-05,
+      "loss": 0.4581,
+      "step": 12347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802952233131744e-05,
+      "loss": 0.404,
+      "step": 12348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2801876424192834e-05,
+      "loss": 0.502,
+      "step": 12349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800800580062202e-05,
+      "loss": 0.509,
+      "step": 12350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2799724700753357e-05,
+      "loss": 0.4593,
+      "step": 12351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798648786279812e-05,
+      "loss": 0.5129,
+      "step": 12352
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797572836655084e-05,
+      "loss": 0.4623,
+      "step": 12353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279649685189268e-05,
+      "loss": 0.469,
+      "step": 12354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279542083200612e-05,
+      "loss": 0.4299,
+      "step": 12355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279434477700892e-05,
+      "loss": 0.4467,
+      "step": 12356
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2793268686914593e-05,
+      "loss": 0.4734,
+      "step": 12357
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792192561736654e-05,
+      "loss": 0.4567,
+      "step": 12358
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791116401488618e-05,
+      "loss": 0.4866,
+      "step": 12359
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790040206184007e-05,
+      "loss": 0.4509,
+      "step": 12360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788963975836334e-05,
+      "loss": 0.4578,
+      "step": 12361
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2787887710459118e-05,
+      "loss": 0.4737,
+      "step": 12362
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786811410065874e-05,
+      "loss": 0.4833,
+      "step": 12363
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785735074670123e-05,
+      "loss": 0.3876,
+      "step": 12364
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784658704285385e-05,
+      "loss": 0.4669,
+      "step": 12365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783582298925177e-05,
+      "loss": 0.4003,
+      "step": 12366
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278250585860302e-05,
+      "loss": 0.4471,
+      "step": 12367
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2781429383332431e-05,
+      "loss": 0.4674,
+      "step": 12368
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780352873126935e-05,
+      "loss": 0.3851,
+      "step": 12369
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779276328000052e-05,
+      "loss": 0.4636,
+      "step": 12370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2778199747965302e-05,
+      "loss": 0.4182,
+      "step": 12371
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2777123133036207e-05,
+      "loss": 0.4767,
+      "step": 12372
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277604648322629e-05,
+      "loss": 0.4221,
+      "step": 12373
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2774969798549073e-05,
+      "loss": 0.4683,
+      "step": 12374
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277389307901808e-05,
+      "loss": 0.4737,
+      "step": 12375
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772816324646837e-05,
+      "loss": 0.4643,
+      "step": 12376
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2771739535448864e-05,
+      "loss": 0.464,
+      "step": 12377
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277066271143769e-05,
+      "loss": 0.3971,
+      "step": 12378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2769585852626832e-05,
+      "loss": 0.4909,
+      "step": 12379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2768508959029822e-05,
+      "loss": 0.4449,
+      "step": 12380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767432030660185e-05,
+      "loss": 0.4268,
+      "step": 12381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2766355067531449e-05,
+      "loss": 0.5025,
+      "step": 12382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765278069657136e-05,
+      "loss": 0.433,
+      "step": 12383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2764201037050776e-05,
+      "loss": 0.4617,
+      "step": 12384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2763123969725894e-05,
+      "loss": 0.498,
+      "step": 12385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762046867696025e-05,
+      "loss": 0.4568,
+      "step": 12386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760969730974692e-05,
+      "loss": 0.4477,
+      "step": 12387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275989255957542e-05,
+      "loss": 0.5199,
+      "step": 12388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758815353511747e-05,
+      "loss": 0.4464,
+      "step": 12389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2757738112797198e-05,
+      "loss": 0.373,
+      "step": 12390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27566608374453e-05,
+      "loss": 0.4926,
+      "step": 12391
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755583527469592e-05,
+      "loss": 0.529,
+      "step": 12392
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27545061828836e-05,
+      "loss": 0.4536,
+      "step": 12393
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753428803700853e-05,
+      "loss": 0.4859,
+      "step": 12394
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275235138993489e-05,
+      "loss": 0.4631,
+      "step": 12395
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751273941599236e-05,
+      "loss": 0.4792,
+      "step": 12396
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2750196458707431e-05,
+      "loss": 0.4043,
+      "step": 12397
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749118941273002e-05,
+      "loss": 0.4399,
+      "step": 12398
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748041389309485e-05,
+      "loss": 0.4442,
+      "step": 12399
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746963802830415e-05,
+      "loss": 0.4504,
+      "step": 12400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745886181849325e-05,
+      "loss": 0.4283,
+      "step": 12401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274480852637975e-05,
+      "loss": 0.4757,
+      "step": 12402
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274373083643523e-05,
+      "loss": 0.492,
+      "step": 12403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742653112029295e-05,
+      "loss": 0.4623,
+      "step": 12404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274157535317548e-05,
+      "loss": 0.456,
+      "step": 12405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740497559887328e-05,
+      "loss": 0.492,
+      "step": 12406
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2739419732178374e-05,
+      "loss": 0.4543,
+      "step": 12407
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738341870062154e-05,
+      "loss": 0.509,
+      "step": 12408
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737263973552207e-05,
+      "loss": 0.4505,
+      "step": 12409
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736186042662072e-05,
+      "loss": 0.4339,
+      "step": 12410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735108077405282e-05,
+      "loss": 0.4333,
+      "step": 12411
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2734030077795384e-05,
+      "loss": 0.4401,
+      "step": 12412
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732952043845917e-05,
+      "loss": 0.4999,
+      "step": 12413
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273187397557042e-05,
+      "loss": 0.407,
+      "step": 12414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730795872982432e-05,
+      "loss": 0.4016,
+      "step": 12415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272971773609549e-05,
+      "loss": 0.4535,
+      "step": 12416
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728639564923146e-05,
+      "loss": 0.3919,
+      "step": 12417
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727561359478934e-05,
+      "loss": 0.4882,
+      "step": 12418
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2726483119776401e-05,
+      "loss": 0.4886,
+      "step": 12419
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2725404845829086e-05,
+      "loss": 0.4585,
+      "step": 12420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724326537650533e-05,
+      "loss": 0.4431,
+      "step": 12421
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723248195254284e-05,
+      "loss": 0.4801,
+      "step": 12422
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272216981865389e-05,
+      "loss": 0.4522,
+      "step": 12423
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272109140786289e-05,
+      "loss": 0.4369,
+      "step": 12424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2720012962894825e-05,
+      "loss": 0.5181,
+      "step": 12425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718934483763248e-05,
+      "loss": 0.4965,
+      "step": 12426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27178559704817e-05,
+      "loss": 0.4953,
+      "step": 12427
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716777423063728e-05,
+      "loss": 0.4939,
+      "step": 12428
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715698841522883e-05,
+      "loss": 0.4698,
+      "step": 12429
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714620225872708e-05,
+      "loss": 0.4705,
+      "step": 12430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713541576126749e-05,
+      "loss": 0.4458,
+      "step": 12431
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712462892298556e-05,
+      "loss": 0.4374,
+      "step": 12432
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2711384174401676e-05,
+      "loss": 0.471,
+      "step": 12433
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2710305422449662e-05,
+      "loss": 0.5196,
+      "step": 12434
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270922663645606e-05,
+      "loss": 0.4586,
+      "step": 12435
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708147816434415e-05,
+      "loss": 0.4777,
+      "step": 12436
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707068962398286e-05,
+      "loss": 0.4831,
+      "step": 12437
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705990074361218e-05,
+      "loss": 0.4646,
+      "step": 12438
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704911152336766e-05,
+      "loss": 0.4531,
+      "step": 12439
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703832196338474e-05,
+      "loss": 0.4214,
+      "step": 12440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27027532063799e-05,
+      "loss": 0.4232,
+      "step": 12441
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701674182474593e-05,
+      "loss": 0.4906,
+      "step": 12442
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270059512463611e-05,
+      "loss": 0.4749,
+      "step": 12443
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699516032877999e-05,
+      "loss": 0.4637,
+      "step": 12444
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698436907213815e-05,
+      "loss": 0.4603,
+      "step": 12445
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697357747657115e-05,
+      "loss": 0.4368,
+      "step": 12446
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2696278554221446e-05,
+      "loss": 0.4622,
+      "step": 12447
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695199326920371e-05,
+      "loss": 0.437,
+      "step": 12448
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694120065767442e-05,
+      "loss": 0.4978,
+      "step": 12449
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269304077077621e-05,
+      "loss": 0.4807,
+      "step": 12450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2691961441960238e-05,
+      "loss": 0.4419,
+      "step": 12451
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269088207933308e-05,
+      "loss": 0.4198,
+      "step": 12452
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268980268290829e-05,
+      "loss": 0.501,
+      "step": 12453
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688723252699431e-05,
+      "loss": 0.468,
+      "step": 12454
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2687643788720055e-05,
+      "loss": 0.4431,
+      "step": 12455
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2686564290983723e-05,
+      "loss": 0.4481,
+      "step": 12456
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685484759503995e-05,
+      "loss": 0.4789,
+      "step": 12457
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684405194294425e-05,
+      "loss": 0.4817,
+      "step": 12458
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2683325595368574e-05,
+      "loss": 0.4165,
+      "step": 12459
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2682245962740008e-05,
+      "loss": 0.4211,
+      "step": 12460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2681166296422282e-05,
+      "loss": 0.4972,
+      "step": 12461
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2680086596428954e-05,
+      "loss": 0.4331,
+      "step": 12462
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267900686277359e-05,
+      "loss": 0.4813,
+      "step": 12463
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677927095469749e-05,
+      "loss": 0.411,
+      "step": 12464
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676847294530993e-05,
+      "loss": 0.4409,
+      "step": 12465
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2675767459970889e-05,
+      "loss": 0.4501,
+      "step": 12466
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267468759180299e-05,
+      "loss": 0.455,
+      "step": 12467
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673607690040868e-05,
+      "loss": 0.48,
+      "step": 12468
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672527754698085e-05,
+      "loss": 0.4437,
+      "step": 12469
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.26714477857882e-05,
+      "loss": 0.4312,
+      "step": 12470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670367783324787e-05,
+      "loss": 0.4474,
+      "step": 12471
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669287747321401e-05,
+      "loss": 0.4507,
+      "step": 12472
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2668207677791611e-05,
+      "loss": 0.4554,
+      "step": 12473
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667127574748985e-05,
+      "loss": 0.4584,
+      "step": 12474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2666047438207087e-05,
+      "loss": 0.5013,
+      "step": 12475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664967268179484e-05,
+      "loss": 0.4476,
+      "step": 12476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663887064679743e-05,
+      "loss": 0.5077,
+      "step": 12477
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266280682772143e-05,
+      "loss": 0.4721,
+      "step": 12478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661726557318113e-05,
+      "loss": 0.4605,
+      "step": 12479
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660646253483364e-05,
+      "loss": 0.4257,
+      "step": 12480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2659565916230747e-05,
+      "loss": 0.4546,
+      "step": 12481
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2658485545573835e-05,
+      "loss": 0.4935,
+      "step": 12482
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657405141526196e-05,
+      "loss": 0.4727,
+      "step": 12483
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656324704101398e-05,
+      "loss": 0.4488,
+      "step": 12484
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2655244233313014e-05,
+      "loss": 0.462,
+      "step": 12485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654163729174612e-05,
+      "loss": 0.4455,
+      "step": 12486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653083191699766e-05,
+      "loss": 0.4176,
+      "step": 12487
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2652002620902046e-05,
+      "loss": 0.4347,
+      "step": 12488
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650922016795027e-05,
+      "loss": 0.4402,
+      "step": 12489
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2649841379392274e-05,
+      "loss": 0.485,
+      "step": 12490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264876070870737e-05,
+      "loss": 0.48,
+      "step": 12491
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647680004753879e-05,
+      "loss": 0.4408,
+      "step": 12492
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646599267545383e-05,
+      "loss": 0.4902,
+      "step": 12493
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264551849709545e-05,
+      "loss": 0.4393,
+      "step": 12494
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644437693417657e-05,
+      "loss": 0.5417,
+      "step": 12495
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2643356856525577e-05,
+      "loss": 0.4426,
+      "step": 12496
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642275986432789e-05,
+      "loss": 0.4864,
+      "step": 12497
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641195083152867e-05,
+      "loss": 0.3952,
+      "step": 12498
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2640114146699382e-05,
+      "loss": 0.4515,
+      "step": 12499
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263903317708592e-05,
+      "loss": 0.4692,
+      "step": 12500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637952174326054e-05,
+      "loss": 0.4792,
+      "step": 12501
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636871138433356e-05,
+      "loss": 0.445,
+      "step": 12502
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635790069421414e-05,
+      "loss": 0.4921,
+      "step": 12503
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634708967303797e-05,
+      "loss": 0.4908,
+      "step": 12504
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263362783209409e-05,
+      "loss": 0.4787,
+      "step": 12505
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263254666380587e-05,
+      "loss": 0.5163,
+      "step": 12506
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631465462452717e-05,
+      "loss": 0.4574,
+      "step": 12507
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630384228048212e-05,
+      "loss": 0.4699,
+      "step": 12508
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629302960605928e-05,
+      "loss": 0.4366,
+      "step": 12509
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628221660139457e-05,
+      "loss": 0.4841,
+      "step": 12510
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2627140326662372e-05,
+      "loss": 0.4382,
+      "step": 12511
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262605896018826e-05,
+      "loss": 0.4967,
+      "step": 12512
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2624977560730697e-05,
+      "loss": 0.4615,
+      "step": 12513
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262389612830327e-05,
+      "loss": 0.4505,
+      "step": 12514
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622814662919562e-05,
+      "loss": 0.5006,
+      "step": 12515
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621733164593154e-05,
+      "loss": 0.4412,
+      "step": 12516
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2620651633337628e-05,
+      "loss": 0.4512,
+      "step": 12517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619570069166576e-05,
+      "loss": 0.3987,
+      "step": 12518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618488472093573e-05,
+      "loss": 0.4488,
+      "step": 12519
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2617406842132211e-05,
+      "loss": 0.5001,
+      "step": 12520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261632517929607e-05,
+      "loss": 0.4769,
+      "step": 12521
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615243483598738e-05,
+      "loss": 0.4998,
+      "step": 12522
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614161755053804e-05,
+      "loss": 0.4484,
+      "step": 12523
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261307999367485e-05,
+      "loss": 0.4643,
+      "step": 12524
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2611998199475468e-05,
+      "loss": 0.467,
+      "step": 12525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610916372469239e-05,
+      "loss": 0.4058,
+      "step": 12526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609834512669755e-05,
+      "loss": 0.3965,
+      "step": 12527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2608752620090604e-05,
+      "loss": 0.469,
+      "step": 12528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607670694745372e-05,
+      "loss": 0.4171,
+      "step": 12529
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606588736647652e-05,
+      "loss": 0.4751,
+      "step": 12530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260550674581103e-05,
+      "loss": 0.4413,
+      "step": 12531
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26044247222491e-05,
+      "loss": 0.5152,
+      "step": 12532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603342665975445e-05,
+      "loss": 0.4403,
+      "step": 12533
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602260577003667e-05,
+      "loss": 0.4777,
+      "step": 12534
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601178455347345e-05,
+      "loss": 0.4198,
+      "step": 12535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600096301020077e-05,
+      "loss": 0.501,
+      "step": 12536
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599014114035455e-05,
+      "loss": 0.4642,
+      "step": 12537
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597931894407068e-05,
+      "loss": 0.4195,
+      "step": 12538
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596849642148515e-05,
+      "loss": 0.4529,
+      "step": 12539
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595767357273383e-05,
+      "loss": 0.4863,
+      "step": 12540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594685039795265e-05,
+      "loss": 0.4381,
+      "step": 12541
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259360268972776e-05,
+      "loss": 0.4107,
+      "step": 12542
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259252030708446e-05,
+      "loss": 0.466,
+      "step": 12543
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259143789187896e-05,
+      "loss": 0.4563,
+      "step": 12544
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590355444124854e-05,
+      "loss": 0.5158,
+      "step": 12545
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258927296383574e-05,
+      "loss": 0.4794,
+      "step": 12546
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.4444,
+      "step": 12547
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587107905706863e-05,
+      "loss": 0.4598,
+      "step": 12548
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586025327894299e-05,
+      "loss": 0.4463,
+      "step": 12549
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2584942717601109e-05,
+      "loss": 0.475,
+      "step": 12550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583860074840898e-05,
+      "loss": 0.4797,
+      "step": 12551
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2582777399627256e-05,
+      "loss": 0.4572,
+      "step": 12552
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581694691973782e-05,
+      "loss": 0.4475,
+      "step": 12553
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580611951894084e-05,
+      "loss": 0.4009,
+      "step": 12554
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579529179401752e-05,
+      "loss": 0.4754,
+      "step": 12555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2578446374510392e-05,
+      "loss": 0.4457,
+      "step": 12556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.25773635372336e-05,
+      "loss": 0.4654,
+      "step": 12557
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576280667584974e-05,
+      "loss": 0.4339,
+      "step": 12558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575197765578122e-05,
+      "loss": 0.4734,
+      "step": 12559
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2574114831226642e-05,
+      "loss": 0.4323,
+      "step": 12560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573031864544137e-05,
+      "loss": 0.4663,
+      "step": 12561
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571948865544206e-05,
+      "loss": 0.4348,
+      "step": 12562
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2570865834240454e-05,
+      "loss": 0.4749,
+      "step": 12563
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256978277064648e-05,
+      "loss": 0.5494,
+      "step": 12564
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2568699674775893e-05,
+      "loss": 0.426,
+      "step": 12565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567616546642294e-05,
+      "loss": 0.4478,
+      "step": 12566
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566533386259285e-05,
+      "loss": 0.447,
+      "step": 12567
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2565450193640477e-05,
+      "loss": 0.4741,
+      "step": 12568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2564366968799468e-05,
+      "loss": 0.428,
+      "step": 12569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563283711749867e-05,
+      "loss": 0.4273,
+      "step": 12570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256220042250528e-05,
+      "loss": 0.4634,
+      "step": 12571
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2561117101079312e-05,
+      "loss": 0.4451,
+      "step": 12572
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.256003374748557e-05,
+      "loss": 0.4321,
+      "step": 12573
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255895036173766e-05,
+      "loss": 0.451,
+      "step": 12574
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557866943849187e-05,
+      "loss": 0.4506,
+      "step": 12575
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556783493833766e-05,
+      "loss": 0.4478,
+      "step": 12576
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555700011705002e-05,
+      "loss": 0.4729,
+      "step": 12577
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25546164974765e-05,
+      "loss": 0.3973,
+      "step": 12578
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255353295116187e-05,
+      "loss": 0.5035,
+      "step": 12579
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2552449372774729e-05,
+      "loss": 0.4321,
+      "step": 12580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551365762328676e-05,
+      "loss": 0.4276,
+      "step": 12581
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2550282119837328e-05,
+      "loss": 0.4911,
+      "step": 12582
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549198445314295e-05,
+      "loss": 0.4788,
+      "step": 12583
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548114738773183e-05,
+      "loss": 0.4727,
+      "step": 12584
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254703100022761e-05,
+      "loss": 0.4919,
+      "step": 12585
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2545947229691185e-05,
+      "loss": 0.4759,
+      "step": 12586
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254486342717752e-05,
+      "loss": 0.4624,
+      "step": 12587
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543779592700227e-05,
+      "loss": 0.4665,
+      "step": 12588
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542695726272923e-05,
+      "loss": 0.4634,
+      "step": 12589
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254161182790921e-05,
+      "loss": 0.4359,
+      "step": 12590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540527897622719e-05,
+      "loss": 0.4613,
+      "step": 12591
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253944393542705e-05,
+      "loss": 0.4076,
+      "step": 12592
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538359941335827e-05,
+      "loss": 0.4084,
+      "step": 12593
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2537275915362659e-05,
+      "loss": 0.3733,
+      "step": 12594
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253619185752116e-05,
+      "loss": 0.5007,
+      "step": 12595
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2535107767824954e-05,
+      "loss": 0.4981,
+      "step": 12596
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253402364628765e-05,
+      "loss": 0.4206,
+      "step": 12597
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532939492922868e-05,
+      "loss": 0.5014,
+      "step": 12598
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531855307744225e-05,
+      "loss": 0.4623,
+      "step": 12599
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253077109076534e-05,
+      "loss": 0.4919,
+      "step": 12600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529686841999821e-05,
+      "loss": 0.4714,
+      "step": 12601
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528602561461298e-05,
+      "loss": 0.5091,
+      "step": 12602
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2527518249163389e-05,
+      "loss": 0.4379,
+      "step": 12603
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2526433905119702e-05,
+      "loss": 0.4727,
+      "step": 12604
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2525349529343866e-05,
+      "loss": 0.5212,
+      "step": 12605
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25242651218495e-05,
+      "loss": 0.4487,
+      "step": 12606
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252318068265022e-05,
+      "loss": 0.4618,
+      "step": 12607
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522096211759654e-05,
+      "loss": 0.4534,
+      "step": 12608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2521011709191418e-05,
+      "loss": 0.4561,
+      "step": 12609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519927174959128e-05,
+      "loss": 0.4645,
+      "step": 12610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2518842609076414e-05,
+      "loss": 0.4763,
+      "step": 12611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2517758011556895e-05,
+      "loss": 0.4746,
+      "step": 12612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25166733824142e-05,
+      "loss": 0.4375,
+      "step": 12613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515588721661942e-05,
+      "loss": 0.3952,
+      "step": 12614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514504029313748e-05,
+      "loss": 0.4626,
+      "step": 12615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513419305383243e-05,
+      "loss": 0.4712,
+      "step": 12616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512334549884054e-05,
+      "loss": 0.4982,
+      "step": 12617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511249762829797e-05,
+      "loss": 0.4178,
+      "step": 12618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2510164944234107e-05,
+      "loss": 0.4275,
+      "step": 12619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509080094110604e-05,
+      "loss": 0.5029,
+      "step": 12620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250799521247291e-05,
+      "loss": 0.4534,
+      "step": 12621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506910299334658e-05,
+      "loss": 0.4571,
+      "step": 12622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2505825354709475e-05,
+      "loss": 0.46,
+      "step": 12623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2504740378610985e-05,
+      "loss": 0.4681,
+      "step": 12624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2503655371052812e-05,
+      "loss": 0.4667,
+      "step": 12625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502570332048592e-05,
+      "loss": 0.4483,
+      "step": 12626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2501485261611943e-05,
+      "loss": 0.4612,
+      "step": 12627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25004001597565e-05,
+      "loss": 0.5274,
+      "step": 12628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499315026495895e-05,
+      "loss": 0.4415,
+      "step": 12629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2498229861843751e-05,
+      "loss": 0.4838,
+      "step": 12630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24971446658137e-05,
+      "loss": 0.468,
+      "step": 12631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249605943841937e-05,
+      "loss": 0.4851,
+      "step": 12632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2494974179674398e-05,
+      "loss": 0.4998,
+      "step": 12633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493888889592407e-05,
+      "loss": 0.4488,
+      "step": 12634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492803568187034e-05,
+      "loss": 0.4986,
+      "step": 12635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491718215471909e-05,
+      "loss": 0.4773,
+      "step": 12636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490632831460663e-05,
+      "loss": 0.5133,
+      "step": 12637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489547416166928e-05,
+      "loss": 0.4626,
+      "step": 12638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488461969604338e-05,
+      "loss": 0.4218,
+      "step": 12639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487376491786529e-05,
+      "loss": 0.4809,
+      "step": 12640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2486290982727128e-05,
+      "loss": 0.4557,
+      "step": 12641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485205442439776e-05,
+      "loss": 0.5196,
+      "step": 12642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484119870938102e-05,
+      "loss": 0.4389,
+      "step": 12643
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483034268235743e-05,
+      "loss": 0.4144,
+      "step": 12644
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248194863434634e-05,
+      "loss": 0.4253,
+      "step": 12645
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248086296928352e-05,
+      "loss": 0.4573,
+      "step": 12646
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247977727306092e-05,
+      "loss": 0.4266,
+      "step": 12647
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247869154569218e-05,
+      "loss": 0.4701,
+      "step": 12648
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477605787190934e-05,
+      "loss": 0.4599,
+      "step": 12649
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476519997570824e-05,
+      "loss": 0.5087,
+      "step": 12650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475434176845482e-05,
+      "loss": 0.4973,
+      "step": 12651
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2474348325028548e-05,
+      "loss": 0.5034,
+      "step": 12652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247326244213366e-05,
+      "loss": 0.5157,
+      "step": 12653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472176528174458e-05,
+      "loss": 0.4549,
+      "step": 12654
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247109058316458e-05,
+      "loss": 0.4662,
+      "step": 12655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470004607117667e-05,
+      "loss": 0.3817,
+      "step": 12656
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2468918600047357e-05,
+      "loss": 0.4202,
+      "step": 12657
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246783256196729e-05,
+      "loss": 0.4156,
+      "step": 12658
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246674649289111e-05,
+      "loss": 0.4652,
+      "step": 12659
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465660392832455e-05,
+      "loss": 0.4465,
+      "step": 12660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2464574261804967e-05,
+      "loss": 0.4804,
+      "step": 12661
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2463488099822288e-05,
+      "loss": 0.4338,
+      "step": 12662
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246240190689806e-05,
+      "loss": 0.3827,
+      "step": 12663
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461315683045927e-05,
+      "loss": 0.4756,
+      "step": 12664
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246022942827953e-05,
+      "loss": 0.4785,
+      "step": 12665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2459143142612518e-05,
+      "loss": 0.471,
+      "step": 12666
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458056826058526e-05,
+      "loss": 0.4563,
+      "step": 12667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2456970478631206e-05,
+      "loss": 0.4345,
+      "step": 12668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455884100344193e-05,
+      "loss": 0.4438,
+      "step": 12669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2454797691211142e-05,
+      "loss": 0.4783,
+      "step": 12670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453711251245697e-05,
+      "loss": 0.4662,
+      "step": 12671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2452624780461498e-05,
+      "loss": 0.4441,
+      "step": 12672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451538278872194e-05,
+      "loss": 0.4036,
+      "step": 12673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450451746491434e-05,
+      "loss": 0.4321,
+      "step": 12674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2449365183332862e-05,
+      "loss": 0.444,
+      "step": 12675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448278589410128e-05,
+      "loss": 0.4865,
+      "step": 12676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447191964736875e-05,
+      "loss": 0.3959,
+      "step": 12677
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2446105309326755e-05,
+      "loss": 0.4651,
+      "step": 12678
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445018623193415e-05,
+      "loss": 0.4718,
+      "step": 12679
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2443931906350503e-05,
+      "loss": 0.4625,
+      "step": 12680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2442845158811671e-05,
+      "loss": 0.5069,
+      "step": 12681
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441758380590567e-05,
+      "loss": 0.526,
+      "step": 12682
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440671571700842e-05,
+      "loss": 0.4394,
+      "step": 12683
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2439584732156143e-05,
+      "loss": 0.4956,
+      "step": 12684
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438497861970124e-05,
+      "loss": 0.4453,
+      "step": 12685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437410961156435e-05,
+      "loss": 0.4881,
+      "step": 12686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2436324029728728e-05,
+      "loss": 0.4806,
+      "step": 12687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435237067700657e-05,
+      "loss": 0.4791,
+      "step": 12688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434150075085866e-05,
+      "loss": 0.451,
+      "step": 12689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243306305189802e-05,
+      "loss": 0.4629,
+      "step": 12690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431975998150763e-05,
+      "loss": 0.4777,
+      "step": 12691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2430888913857752e-05,
+      "loss": 0.4859,
+      "step": 12692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242980179903264e-05,
+      "loss": 0.4778,
+      "step": 12693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428714653689083e-05,
+      "loss": 0.5183,
+      "step": 12694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427627477840728e-05,
+      "loss": 0.4392,
+      "step": 12695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426540271501242e-05,
+      "loss": 0.4726,
+      "step": 12696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425453034684274e-05,
+      "loss": 0.4752,
+      "step": 12697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424365767403477e-05,
+      "loss": 0.4677,
+      "step": 12698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423278469672514e-05,
+      "loss": 0.4665,
+      "step": 12699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422191141505033e-05,
+      "loss": 0.5265,
+      "step": 12700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421103782914696e-05,
+      "loss": 0.45,
+      "step": 12701
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2420016393915164e-05,
+      "loss": 0.4623,
+      "step": 12702
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418928974520088e-05,
+      "loss": 0.4773,
+      "step": 12703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417841524743128e-05,
+      "loss": 0.427,
+      "step": 12704
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2416754044597945e-05,
+      "loss": 0.4061,
+      "step": 12705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415666534098194e-05,
+      "loss": 0.4674,
+      "step": 12706
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414578993257535e-05,
+      "loss": 0.4737,
+      "step": 12707
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2413491422089634e-05,
+      "loss": 0.4424,
+      "step": 12708
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412403820608142e-05,
+      "loss": 0.5204,
+      "step": 12709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411316188826724e-05,
+      "loss": 0.4571,
+      "step": 12710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410228526759038e-05,
+      "loss": 0.4306,
+      "step": 12711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409140834418745e-05,
+      "loss": 0.4403,
+      "step": 12712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2408053111819513e-05,
+      "loss": 0.5111,
+      "step": 12713
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406965358974997e-05,
+      "loss": 0.455,
+      "step": 12714
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240587757589886e-05,
+      "loss": 0.4493,
+      "step": 12715
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404789762604766e-05,
+      "loss": 0.4712,
+      "step": 12716
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403701919106378e-05,
+      "loss": 0.416,
+      "step": 12717
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240261404541736e-05,
+      "loss": 0.4679,
+      "step": 12718
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2401526141551377e-05,
+      "loss": 0.4671,
+      "step": 12719
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2400438207522087e-05,
+      "loss": 0.5158,
+      "step": 12720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239935024334316e-05,
+      "loss": 0.4751,
+      "step": 12721
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2398262249028261e-05,
+      "loss": 0.4801,
+      "step": 12722
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397174224591053e-05,
+      "loss": 0.4742,
+      "step": 12723
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396086170045203e-05,
+      "loss": 0.4554,
+      "step": 12724
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394998085404377e-05,
+      "loss": 0.4514,
+      "step": 12725
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393909970682239e-05,
+      "loss": 0.4783,
+      "step": 12726
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2392821825892459e-05,
+      "loss": 0.4809,
+      "step": 12727
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.23917336510487e-05,
+      "loss": 0.4832,
+      "step": 12728
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390645446164638e-05,
+      "loss": 0.4656,
+      "step": 12729
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238955721125393e-05,
+      "loss": 0.4811,
+      "step": 12730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388468946330255e-05,
+      "loss": 0.4762,
+      "step": 12731
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238738065140727e-05,
+      "loss": 0.4377,
+      "step": 12732
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386292326498655e-05,
+      "loss": 0.4094,
+      "step": 12733
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385203971618075e-05,
+      "loss": 0.5077,
+      "step": 12734
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384115586779196e-05,
+      "loss": 0.4467,
+      "step": 12735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383027171995693e-05,
+      "loss": 0.4804,
+      "step": 12736
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2381938727281238e-05,
+      "loss": 0.4433,
+      "step": 12737
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380850252649495e-05,
+      "loss": 0.481,
+      "step": 12738
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237976174811414e-05,
+      "loss": 0.4182,
+      "step": 12739
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2378673213688845e-05,
+      "loss": 0.4631,
+      "step": 12740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377584649387285e-05,
+      "loss": 0.4658,
+      "step": 12741
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376496055223124e-05,
+      "loss": 0.4798,
+      "step": 12742
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2375407431210038e-05,
+      "loss": 0.4312,
+      "step": 12743
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2374318777361704e-05,
+      "loss": 0.4844,
+      "step": 12744
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373230093691797e-05,
+      "loss": 0.4394,
+      "step": 12745
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2372141380213983e-05,
+      "loss": 0.4574,
+      "step": 12746
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237105263694194e-05,
+      "loss": 0.457,
+      "step": 12747
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369963863889346e-05,
+      "loss": 0.4798,
+      "step": 12748
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368875061069869e-05,
+      "loss": 0.3853,
+      "step": 12749
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367786228497194e-05,
+      "loss": 0.4873,
+      "step": 12750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2366697366184987e-05,
+      "loss": 0.4483,
+      "step": 12751
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236560847414693e-05,
+      "loss": 0.4749,
+      "step": 12752
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364519552396699e-05,
+      "loss": 0.4336,
+      "step": 12753
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363430600947968e-05,
+      "loss": 0.5201,
+      "step": 12754
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2362341619814417e-05,
+      "loss": 0.4133,
+      "step": 12755
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361252609009725e-05,
+      "loss": 0.4506,
+      "step": 12756
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360163568547563e-05,
+      "loss": 0.4491,
+      "step": 12757
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2359074498441619e-05,
+      "loss": 0.4754,
+      "step": 12758
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357985398705566e-05,
+      "loss": 0.5279,
+      "step": 12759
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2356896269353083e-05,
+      "loss": 0.4398,
+      "step": 12760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2355807110397853e-05,
+      "loss": 0.45,
+      "step": 12761
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2354717921853551e-05,
+      "loss": 0.4705,
+      "step": 12762
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353628703733863e-05,
+      "loss": 0.4663,
+      "step": 12763
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2352539456052463e-05,
+      "loss": 0.4682,
+      "step": 12764
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.235145017882304e-05,
+      "loss": 0.4964,
+      "step": 12765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2350360872059268e-05,
+      "loss": 0.4302,
+      "step": 12766
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2349271535774834e-05,
+      "loss": 0.4098,
+      "step": 12767
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2348182169983418e-05,
+      "loss": 0.389,
+      "step": 12768
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2347092774698699e-05,
+      "loss": 0.487,
+      "step": 12769
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346003349934366e-05,
+      "loss": 0.4599,
+      "step": 12770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344913895704099e-05,
+      "loss": 0.4611,
+      "step": 12771
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2343824412021582e-05,
+      "loss": 0.4469,
+      "step": 12772
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342734898900499e-05,
+      "loss": 0.4406,
+      "step": 12773
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2341645356354537e-05,
+      "loss": 0.4961,
+      "step": 12774
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340555784397372e-05,
+      "loss": 0.464,
+      "step": 12775
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339466183042701e-05,
+      "loss": 0.4527,
+      "step": 12776
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23383765523042e-05,
+      "loss": 0.5009,
+      "step": 12777
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337286892195563e-05,
+      "loss": 0.427,
+      "step": 12778
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336197202730469e-05,
+      "loss": 0.4604,
+      "step": 12779
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2335107483922605e-05,
+      "loss": 0.4653,
+      "step": 12780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334017735785663e-05,
+      "loss": 0.4354,
+      "step": 12781
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332927958333331e-05,
+      "loss": 0.4463,
+      "step": 12782
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2331838151579287e-05,
+      "loss": 0.496,
+      "step": 12783
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330748315537229e-05,
+      "loss": 0.4782,
+      "step": 12784
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232965845022084e-05,
+      "loss": 0.4439,
+      "step": 12785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232856855564381e-05,
+      "loss": 0.4579,
+      "step": 12786
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232747863181983e-05,
+      "loss": 0.4444,
+      "step": 12787
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326388678762588e-05,
+      "loss": 0.4523,
+      "step": 12788
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232529869648577e-05,
+      "loss": 0.4185,
+      "step": 12789
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2324208685003074e-05,
+      "loss": 0.4566,
+      "step": 12790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323118644328186e-05,
+      "loss": 0.469,
+      "step": 12791
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322028574474799e-05,
+      "loss": 0.4401,
+      "step": 12792
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2320938475456603e-05,
+      "loss": 0.4241,
+      "step": 12793
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319848347287286e-05,
+      "loss": 0.4447,
+      "step": 12794
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318758189980547e-05,
+      "loss": 0.4937,
+      "step": 12795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317668003550074e-05,
+      "loss": 0.3914,
+      "step": 12796
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2316577788009562e-05,
+      "loss": 0.4839,
+      "step": 12797
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315487543372703e-05,
+      "loss": 0.5053,
+      "step": 12798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314397269653192e-05,
+      "loss": 0.4483,
+      "step": 12799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313306966864717e-05,
+      "loss": 0.4877,
+      "step": 12800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312216635020981e-05,
+      "loss": 0.4862,
+      "step": 12801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311126274135673e-05,
+      "loss": 0.4773,
+      "step": 12802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310035884222493e-05,
+      "loss": 0.4102,
+      "step": 12803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.230894546529513e-05,
+      "loss": 0.4557,
+      "step": 12804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307855017367282e-05,
+      "loss": 0.4265,
+      "step": 12805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306764540452647e-05,
+      "loss": 0.4963,
+      "step": 12806
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2305674034564921e-05,
+      "loss": 0.5084,
+      "step": 12807
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304583499717801e-05,
+      "loss": 0.4733,
+      "step": 12808
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303492935924979e-05,
+      "loss": 0.4547,
+      "step": 12809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302402343200159e-05,
+      "loss": 0.464,
+      "step": 12810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2301311721557036e-05,
+      "loss": 0.4613,
+      "step": 12811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300221071009308e-05,
+      "loss": 0.4458,
+      "step": 12812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229913039157068e-05,
+      "loss": 0.4825,
+      "step": 12813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2298039683254839e-05,
+      "loss": 0.5182,
+      "step": 12814
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296948946075494e-05,
+      "loss": 0.4341,
+      "step": 12815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229585818004634e-05,
+      "loss": 0.4872,
+      "step": 12816
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2294767385181076e-05,
+      "loss": 0.4416,
+      "step": 12817
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293676561493412e-05,
+      "loss": 0.4671,
+      "step": 12818
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229258570899704e-05,
+      "loss": 0.465,
+      "step": 12819
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291494827705659e-05,
+      "loss": 0.4642,
+      "step": 12820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290403917632977e-05,
+      "loss": 0.4754,
+      "step": 12821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289312978792695e-05,
+      "loss": 0.4468,
+      "step": 12822
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2288222011198511e-05,
+      "loss": 0.5344,
+      "step": 12823
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287131014864133e-05,
+      "loss": 0.4632,
+      "step": 12824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2286039989803262e-05,
+      "loss": 0.4638,
+      "step": 12825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2284948936029597e-05,
+      "loss": 0.4478,
+      "step": 12826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.228385785355685e-05,
+      "loss": 0.4913,
+      "step": 12827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2282766742398716e-05,
+      "loss": 0.4762,
+      "step": 12828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281675602568906e-05,
+      "loss": 0.4649,
+      "step": 12829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280584434081123e-05,
+      "loss": 0.4644,
+      "step": 12830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279493236949069e-05,
+      "loss": 0.5003,
+      "step": 12831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278402011186454e-05,
+      "loss": 0.4569,
+      "step": 12832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2277310756806983e-05,
+      "loss": 0.4392,
+      "step": 12833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2276219473824362e-05,
+      "loss": 0.4621,
+      "step": 12834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275128162252297e-05,
+      "loss": 0.4503,
+      "step": 12835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2274036822104493e-05,
+      "loss": 0.4568,
+      "step": 12836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272945453394658e-05,
+      "loss": 0.454,
+      "step": 12837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271854056136503e-05,
+      "loss": 0.513,
+      "step": 12838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270762630343734e-05,
+      "loss": 0.487,
+      "step": 12839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269671176030058e-05,
+      "loss": 0.4639,
+      "step": 12840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2268579693209186e-05,
+      "loss": 0.4528,
+      "step": 12841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267488181894826e-05,
+      "loss": 0.468,
+      "step": 12842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266396642100684e-05,
+      "loss": 0.4414,
+      "step": 12843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265305073840476e-05,
+      "loss": 0.447,
+      "step": 12844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264213477127914e-05,
+      "loss": 0.4571,
+      "step": 12845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2263121851976698e-05,
+      "loss": 0.418,
+      "step": 12846
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2262030198400545e-05,
+      "loss": 0.4289,
+      "step": 12847
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260938516413168e-05,
+      "loss": 0.5036,
+      "step": 12848
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2259846806028275e-05,
+      "loss": 0.4273,
+      "step": 12849
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2258755067259582e-05,
+      "loss": 0.4463,
+      "step": 12850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2257663300120797e-05,
+      "loss": 0.5144,
+      "step": 12851
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2256571504625634e-05,
+      "loss": 0.4648,
+      "step": 12852
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2255479680787809e-05,
+      "loss": 0.412,
+      "step": 12853
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254387828621028e-05,
+      "loss": 0.482,
+      "step": 12854
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2253295948139016e-05,
+      "loss": 0.4435,
+      "step": 12855
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252204039355477e-05,
+      "loss": 0.479,
+      "step": 12856
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2251112102284129e-05,
+      "loss": 0.4822,
+      "step": 12857
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2250020136938688e-05,
+      "loss": 0.4354,
+      "step": 12858
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2248928143332868e-05,
+      "loss": 0.5057,
+      "step": 12859
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2247836121480386e-05,
+      "loss": 0.4592,
+      "step": 12860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246744071394953e-05,
+      "loss": 0.4799,
+      "step": 12861
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224565199309029e-05,
+      "loss": 0.5096,
+      "step": 12862
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244559886580112e-05,
+      "loss": 0.426,
+      "step": 12863
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2243467751878137e-05,
+      "loss": 0.4392,
+      "step": 12864
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242375588998082e-05,
+      "loss": 0.4145,
+      "step": 12865
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241283397953662e-05,
+      "loss": 0.4622,
+      "step": 12866
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240191178758598e-05,
+      "loss": 0.4274,
+      "step": 12867
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2239098931426608e-05,
+      "loss": 0.4555,
+      "step": 12868
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238006655971408e-05,
+      "loss": 0.4732,
+      "step": 12869
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223691435240672e-05,
+      "loss": 0.4972,
+      "step": 12870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2235822020746264e-05,
+      "loss": 0.4429,
+      "step": 12871
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234729661003757e-05,
+      "loss": 0.4728,
+      "step": 12872
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233637273192922e-05,
+      "loss": 0.4503,
+      "step": 12873
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232544857327473e-05,
+      "loss": 0.4542,
+      "step": 12874
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2231452413421139e-05,
+      "loss": 0.4465,
+      "step": 12875
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2230359941487638e-05,
+      "loss": 0.4425,
+      "step": 12876
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222926744154069e-05,
+      "loss": 0.4022,
+      "step": 12877
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222817491359402e-05,
+      "loss": 0.4448,
+      "step": 12878
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227082357661349e-05,
+      "loss": 0.4138,
+      "step": 12879
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225989773756392e-05,
+      "loss": 0.4737,
+      "step": 12880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2224897161892886e-05,
+      "loss": 0.4191,
+      "step": 12881
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223804522084546e-05,
+      "loss": 0.4613,
+      "step": 12882
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2222711854345096e-05,
+      "loss": 0.4486,
+      "step": 12883
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221619158688259e-05,
+      "loss": 0.4878,
+      "step": 12884
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222052643512776e-05,
+      "loss": 0.4694,
+      "step": 12885
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2219433683677326e-05,
+      "loss": 0.4558,
+      "step": 12886
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218340904350685e-05,
+      "loss": 0.4329,
+      "step": 12887
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2217248097161552e-05,
+      "loss": 0.5133,
+      "step": 12888
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2216155262123661e-05,
+      "loss": 0.4662,
+      "step": 12889
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215062399250735e-05,
+      "loss": 0.5098,
+      "step": 12890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22139695085565e-05,
+      "loss": 0.4678,
+      "step": 12891
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221287659005469e-05,
+      "loss": 0.4585,
+      "step": 12892
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211783643759022e-05,
+      "loss": 0.4805,
+      "step": 12893
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2210690669683226e-05,
+      "loss": 0.4161,
+      "step": 12894
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209597667841031e-05,
+      "loss": 0.4927,
+      "step": 12895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208504638246167e-05,
+      "loss": 0.4997,
+      "step": 12896
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2207411580912362e-05,
+      "loss": 0.4368,
+      "step": 12897
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206318495853346e-05,
+      "loss": 0.4846,
+      "step": 12898
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205225383082844e-05,
+      "loss": 0.4947,
+      "step": 12899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204132242614583e-05,
+      "loss": 0.4635,
+      "step": 12900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203039074462304e-05,
+      "loss": 0.4416,
+      "step": 12901
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220194587863973e-05,
+      "loss": 0.4868,
+      "step": 12902
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2200852655160592e-05,
+      "loss": 0.5472,
+      "step": 12903
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199759404038618e-05,
+      "loss": 0.4498,
+      "step": 12904
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198666125287544e-05,
+      "loss": 0.4716,
+      "step": 12905
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197572818921101e-05,
+      "loss": 0.4709,
+      "step": 12906
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219647948495302e-05,
+      "loss": 0.4712,
+      "step": 12907
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2195386123397037e-05,
+      "loss": 0.4755,
+      "step": 12908
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219429273426688e-05,
+      "loss": 0.4576,
+      "step": 12909
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219319931757628e-05,
+      "loss": 0.4754,
+      "step": 12910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2192105873338975e-05,
+      "loss": 0.4624,
+      "step": 12911
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191012401568698e-05,
+      "loss": 0.5141,
+      "step": 12912
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189918902279183e-05,
+      "loss": 0.4806,
+      "step": 12913
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188825375484165e-05,
+      "loss": 0.4955,
+      "step": 12914
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2187731821197377e-05,
+      "loss": 0.4401,
+      "step": 12915
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2186638239432553e-05,
+      "loss": 0.4179,
+      "step": 12916
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185544630203432e-05,
+      "loss": 0.4406,
+      "step": 12917
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2184450993523746e-05,
+      "loss": 0.452,
+      "step": 12918
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183357329407239e-05,
+      "loss": 0.4585,
+      "step": 12919
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2182263637867635e-05,
+      "loss": 0.4457,
+      "step": 12920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218116991891868e-05,
+      "loss": 0.5021,
+      "step": 12921
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2180076172574106e-05,
+      "loss": 0.4934,
+      "step": 12922
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178982398847656e-05,
+      "loss": 0.4156,
+      "step": 12923
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177888597753066e-05,
+      "loss": 0.4514,
+      "step": 12924
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176794769304073e-05,
+      "loss": 0.4327,
+      "step": 12925
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175700913514415e-05,
+      "loss": 0.4591,
+      "step": 12926
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2174607030397832e-05,
+      "loss": 0.4627,
+      "step": 12927
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173513119968058e-05,
+      "loss": 0.4812,
+      "step": 12928
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172419182238846e-05,
+      "loss": 0.5185,
+      "step": 12929
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2171325217223923e-05,
+      "loss": 0.429,
+      "step": 12930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170231224937032e-05,
+      "loss": 0.4888,
+      "step": 12931
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2169137205391918e-05,
+      "loss": 0.466,
+      "step": 12932
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2168043158602319e-05,
+      "loss": 0.418,
+      "step": 12933
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166949084581975e-05,
+      "loss": 0.4505,
+      "step": 12934
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165854983344632e-05,
+      "loss": 0.4763,
+      "step": 12935
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164760854904028e-05,
+      "loss": 0.4592,
+      "step": 12936
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2163666699273901e-05,
+      "loss": 0.5028,
+      "step": 12937
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162572516468005e-05,
+      "loss": 0.4984,
+      "step": 12938
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2161478306500077e-05,
+      "loss": 0.4752,
+      "step": 12939
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216038406938386e-05,
+      "loss": 0.478,
+      "step": 12940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2159289805133096e-05,
+      "loss": 0.4508,
+      "step": 12941
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215819551376153e-05,
+      "loss": 0.4355,
+      "step": 12942
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215710119528291e-05,
+      "loss": 0.4343,
+      "step": 12943
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156006849710979e-05,
+      "loss": 0.4346,
+      "step": 12944
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2154912477059482e-05,
+      "loss": 0.4102,
+      "step": 12945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2153818077342161e-05,
+      "loss": 0.4312,
+      "step": 12946
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152723650572767e-05,
+      "loss": 0.4148,
+      "step": 12947
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151629196765039e-05,
+      "loss": 0.4645,
+      "step": 12948
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215053471593273e-05,
+      "loss": 0.4772,
+      "step": 12949
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2149440208089587e-05,
+      "loss": 0.4901,
+      "step": 12950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214834567324935e-05,
+      "loss": 0.4434,
+      "step": 12951
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147251111425775e-05,
+      "loss": 0.502,
+      "step": 12952
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146156522632603e-05,
+      "loss": 0.4317,
+      "step": 12953
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2145061906883585e-05,
+      "loss": 0.4675,
+      "step": 12954
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143967264192474e-05,
+      "loss": 0.4863,
+      "step": 12955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2142872594573009e-05,
+      "loss": 0.465,
+      "step": 12956
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141777898038944e-05,
+      "loss": 0.4494,
+      "step": 12957
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2140683174604032e-05,
+      "loss": 0.4378,
+      "step": 12958
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2139588424282016e-05,
+      "loss": 0.4226,
+      "step": 12959
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138493647086653e-05,
+      "loss": 0.4728,
+      "step": 12960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137398843031689e-05,
+      "loss": 0.4576,
+      "step": 12961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2136304012130872e-05,
+      "loss": 0.4563,
+      "step": 12962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135209154397962e-05,
+      "loss": 0.452,
+      "step": 12963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134114269846704e-05,
+      "loss": 0.487,
+      "step": 12964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213301935849085e-05,
+      "loss": 0.411,
+      "step": 12965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131924420344154e-05,
+      "loss": 0.4628,
+      "step": 12966
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213082945542037e-05,
+      "loss": 0.4929,
+      "step": 12967
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2129734463733245e-05,
+      "loss": 0.4763,
+      "step": 12968
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2128639445296538e-05,
+      "loss": 0.463,
+      "step": 12969
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127544400124001e-05,
+      "loss": 0.3855,
+      "step": 12970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126449328229386e-05,
+      "loss": 0.4364,
+      "step": 12971
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212535422962645e-05,
+      "loss": 0.4669,
+      "step": 12972
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124259104328944e-05,
+      "loss": 0.4288,
+      "step": 12973
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123163952350623e-05,
+      "loss": 0.4796,
+      "step": 12974
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2122068773705245e-05,
+      "loss": 0.4943,
+      "step": 12975
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120973568406566e-05,
+      "loss": 0.4947,
+      "step": 12976
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211987833646834e-05,
+      "loss": 0.4342,
+      "step": 12977
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2118783077904324e-05,
+      "loss": 0.4084,
+      "step": 12978
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211768779272827e-05,
+      "loss": 0.4859,
+      "step": 12979
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116592480953942e-05,
+      "loss": 0.4799,
+      "step": 12980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2115497142595094e-05,
+      "loss": 0.4389,
+      "step": 12981
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114401777665484e-05,
+      "loss": 0.4717,
+      "step": 12982
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113306386178866e-05,
+      "loss": 0.4755,
+      "step": 12983
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2112210968149006e-05,
+      "loss": 0.4248,
+      "step": 12984
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111115523589651e-05,
+      "loss": 0.4641,
+      "step": 12985
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2110020052514573e-05,
+      "loss": 0.4401,
+      "step": 12986
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108924554937524e-05,
+      "loss": 0.4429,
+      "step": 12987
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107829030872263e-05,
+      "loss": 0.4351,
+      "step": 12988
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2106733480332551e-05,
+      "loss": 0.4953,
+      "step": 12989
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210563790333215e-05,
+      "loss": 0.4208,
+      "step": 12990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104542299884816e-05,
+      "loss": 0.4278,
+      "step": 12991
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103446670004319e-05,
+      "loss": 0.4567,
+      "step": 12992
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2102351013704409e-05,
+      "loss": 0.457,
+      "step": 12993
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2101255330998853e-05,
+      "loss": 0.4555,
+      "step": 12994
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100159621901413e-05,
+      "loss": 0.4734,
+      "step": 12995
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209906388642585e-05,
+      "loss": 0.4551,
+      "step": 12996
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209796812458593e-05,
+      "loss": 0.4749,
+      "step": 12997
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209687233639541e-05,
+      "loss": 0.4752,
+      "step": 12998
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095776521868054e-05,
+      "loss": 0.4451,
+      "step": 12999
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2094680681017632e-05,
+      "loss": 0.46,
+      "step": 13000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20935848138579e-05,
+      "loss": 0.4363,
+      "step": 13001
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209248892040263e-05,
+      "loss": 0.4379,
+      "step": 13002
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2091393000665578e-05,
+      "loss": 0.4402,
+      "step": 13003
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090297054660514e-05,
+      "loss": 0.4468,
+      "step": 13004
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20892010824012e-05,
+      "loss": 0.4704,
+      "step": 13005
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088105083901404e-05,
+      "loss": 0.397,
+      "step": 13006
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087009059174895e-05,
+      "loss": 0.4382,
+      "step": 13007
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085913008235433e-05,
+      "loss": 0.4432,
+      "step": 13008
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2084816931096785e-05,
+      "loss": 0.4756,
+      "step": 13009
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208372082777272e-05,
+      "loss": 0.4585,
+      "step": 13010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082624698277005e-05,
+      "loss": 0.4168,
+      "step": 13011
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2081528542623408e-05,
+      "loss": 0.4869,
+      "step": 13012
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080432360825696e-05,
+      "loss": 0.4568,
+      "step": 13013
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079336152897636e-05,
+      "loss": 0.4652,
+      "step": 13014
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2078239918852999e-05,
+      "loss": 0.4674,
+      "step": 13015
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207714365870555e-05,
+      "loss": 0.3872,
+      "step": 13016
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2076047372469062e-05,
+      "loss": 0.4446,
+      "step": 13017
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074951060157303e-05,
+      "loss": 0.4661,
+      "step": 13018
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073854721784043e-05,
+      "loss": 0.4452,
+      "step": 13019
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207275835736305e-05,
+      "loss": 0.5297,
+      "step": 13020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2071661966908098e-05,
+      "loss": 0.4258,
+      "step": 13021
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070565550432951e-05,
+      "loss": 0.4511,
+      "step": 13022
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069469107951388e-05,
+      "loss": 0.4809,
+      "step": 13023
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206837263947718e-05,
+      "loss": 0.4556,
+      "step": 13024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2067276145024089e-05,
+      "loss": 0.4895,
+      "step": 13025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2066179624605897e-05,
+      "loss": 0.4126,
+      "step": 13026
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065083078236375e-05,
+      "loss": 0.4978,
+      "step": 13027
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206398650592929e-05,
+      "loss": 0.4645,
+      "step": 13028
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062889907698425e-05,
+      "loss": 0.4819,
+      "step": 13029
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061793283557543e-05,
+      "loss": 0.4863,
+      "step": 13030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206069663352042e-05,
+      "loss": 0.4752,
+      "step": 13031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059599957600838e-05,
+      "loss": 0.4657,
+      "step": 13032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058503255812559e-05,
+      "loss": 0.4508,
+      "step": 13033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2057406528169368e-05,
+      "loss": 0.4923,
+      "step": 13034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056309774685036e-05,
+      "loss": 0.4947,
+      "step": 13035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2055212995373333e-05,
+      "loss": 0.4574,
+      "step": 13036
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2054116190248044e-05,
+      "loss": 0.4715,
+      "step": 13037
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205301935932294e-05,
+      "loss": 0.4692,
+      "step": 13038
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051922502611798e-05,
+      "loss": 0.5047,
+      "step": 13039
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050825620128394e-05,
+      "loss": 0.4869,
+      "step": 13040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049728711886505e-05,
+      "loss": 0.4561,
+      "step": 13041
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048631777899909e-05,
+      "loss": 0.4701,
+      "step": 13042
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047534818182382e-05,
+      "loss": 0.47,
+      "step": 13043
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2046437832747704e-05,
+      "loss": 0.4888,
+      "step": 13044
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045340821609652e-05,
+      "loss": 0.4593,
+      "step": 13045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044243784782006e-05,
+      "loss": 0.4728,
+      "step": 13046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043146722278542e-05,
+      "loss": 0.5064,
+      "step": 13047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204204963411304e-05,
+      "loss": 0.4661,
+      "step": 13048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2040952520299282e-05,
+      "loss": 0.414,
+      "step": 13049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2039855380851046e-05,
+      "loss": 0.4898,
+      "step": 13050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2038758215782112e-05,
+      "loss": 0.4108,
+      "step": 13051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203766102510626e-05,
+      "loss": 0.4605,
+      "step": 13052
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203656380883727e-05,
+      "loss": 0.4591,
+      "step": 13053
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035466566988927e-05,
+      "loss": 0.4715,
+      "step": 13054
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2034369299575008e-05,
+      "loss": 0.4607,
+      "step": 13055
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.20332720066093e-05,
+      "loss": 0.4861,
+      "step": 13056
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203217468810558e-05,
+      "loss": 0.4182,
+      "step": 13057
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203107734407763e-05,
+      "loss": 0.4928,
+      "step": 13058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2029979974539233e-05,
+      "loss": 0.4751,
+      "step": 13059
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202888257950418e-05,
+      "loss": 0.4833,
+      "step": 13060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027785158986247e-05,
+      "loss": 0.4768,
+      "step": 13061
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2026687712999217e-05,
+      "loss": 0.53,
+      "step": 13062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025590241556874e-05,
+      "loss": 0.4309,
+      "step": 13063
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024492744673008e-05,
+      "loss": 0.4737,
+      "step": 13064
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2023395222361396e-05,
+      "loss": 0.4837,
+      "step": 13065
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202229767463583e-05,
+      "loss": 0.5039,
+      "step": 13066
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021200101510093e-05,
+      "loss": 0.4361,
+      "step": 13067
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020102502997964e-05,
+      "loss": 0.4624,
+      "step": 13068
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2019004879113238e-05,
+      "loss": 0.451,
+      "step": 13069
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017907229869698e-05,
+      "loss": 0.4671,
+      "step": 13070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2016809555281128e-05,
+      "loss": 0.4524,
+      "step": 13071
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.201571185536132e-05,
+      "loss": 0.4453,
+      "step": 13072
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014614130124054e-05,
+      "loss": 0.4368,
+      "step": 13073
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2013516379583126e-05,
+      "loss": 0.4733,
+      "step": 13074
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012418603752316e-05,
+      "loss": 0.4972,
+      "step": 13075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011320802645418e-05,
+      "loss": 0.4205,
+      "step": 13076
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010222976276218e-05,
+      "loss": 0.5027,
+      "step": 13077
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009125124658505e-05,
+      "loss": 0.4875,
+      "step": 13078
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008027247806067e-05,
+      "loss": 0.4337,
+      "step": 13079
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006929345732694e-05,
+      "loss": 0.4869,
+      "step": 13080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2005831418452176e-05,
+      "loss": 0.448,
+      "step": 13081
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004733465978304e-05,
+      "loss": 0.413,
+      "step": 13082
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2003635488324867e-05,
+      "loss": 0.5022,
+      "step": 13083
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2002537485505657e-05,
+      "loss": 0.47,
+      "step": 13084
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200143945753446e-05,
+      "loss": 0.4588,
+      "step": 13085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2000341404425076e-05,
+      "loss": 0.5067,
+      "step": 13086
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199924332619129e-05,
+      "loss": 0.5063,
+      "step": 13087
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998145222846897e-05,
+      "loss": 0.4789,
+      "step": 13088
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997047094405687e-05,
+      "loss": 0.4507,
+      "step": 13089
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1995948940881453e-05,
+      "loss": 0.4281,
+      "step": 13090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994850762287988e-05,
+      "loss": 0.4508,
+      "step": 13091
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1993752558639092e-05,
+      "loss": 0.3945,
+      "step": 13092
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1992654329948542e-05,
+      "loss": 0.4427,
+      "step": 13093
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991556076230148e-05,
+      "loss": 0.4349,
+      "step": 13094
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990457797497699e-05,
+      "loss": 0.4488,
+      "step": 13095
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1989359493764985e-05,
+      "loss": 0.4223,
+      "step": 13096
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198826116504581e-05,
+      "loss": 0.4769,
+      "step": 13097
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987162811353958e-05,
+      "loss": 0.4546,
+      "step": 13098
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198606443270323e-05,
+      "loss": 0.4036,
+      "step": 13099
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984966029107423e-05,
+      "loss": 0.4463,
+      "step": 13100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1983867600580333e-05,
+      "loss": 0.5101,
+      "step": 13101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982769147135754e-05,
+      "loss": 0.4536,
+      "step": 13102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981670668787485e-05,
+      "loss": 0.4186,
+      "step": 13103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980572165549315e-05,
+      "loss": 0.4611,
+      "step": 13104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1979473637435053e-05,
+      "loss": 0.4102,
+      "step": 13105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197837508445849e-05,
+      "loss": 0.5291,
+      "step": 13106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977276506633425e-05,
+      "loss": 0.492,
+      "step": 13107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976177903973657e-05,
+      "loss": 0.466,
+      "step": 13108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1975079276492985e-05,
+      "loss": 0.4056,
+      "step": 13109
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973980624205204e-05,
+      "loss": 0.4293,
+      "step": 13110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1972881947124118e-05,
+      "loss": 0.4793,
+      "step": 13111
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971783245263523e-05,
+      "loss": 0.4194,
+      "step": 13112
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970684518637218e-05,
+      "loss": 0.4766,
+      "step": 13113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969585767259009e-05,
+      "loss": 0.4506,
+      "step": 13114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968486991142692e-05,
+      "loss": 0.4058,
+      "step": 13115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967388190302064e-05,
+      "loss": 0.4685,
+      "step": 13116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1966289364750933e-05,
+      "loss": 0.4727,
+      "step": 13117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965190514503097e-05,
+      "loss": 0.4618,
+      "step": 13118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1964091639572362e-05,
+      "loss": 0.4111,
+      "step": 13119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1962992739972521e-05,
+      "loss": 0.4581,
+      "step": 13120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1961893815717384e-05,
+      "loss": 0.4814,
+      "step": 13121
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196079486682075e-05,
+      "loss": 0.4795,
+      "step": 13122
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959695893296421e-05,
+      "loss": 0.41,
+      "step": 13123
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958596895158203e-05,
+      "loss": 0.4597,
+      "step": 13124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957497872419899e-05,
+      "loss": 0.4965,
+      "step": 13125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956398825095313e-05,
+      "loss": 0.4597,
+      "step": 13126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1955299753198244e-05,
+      "loss": 0.4468,
+      "step": 13127
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954200656742503e-05,
+      "loss": 0.4697,
+      "step": 13128
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1953101535741895e-05,
+      "loss": 0.4722,
+      "step": 13129
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952002390210217e-05,
+      "loss": 0.447,
+      "step": 13130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.5045,
+      "step": 13131
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949804025608897e-05,
+      "loss": 0.4626,
+      "step": 13132
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194870480656686e-05,
+      "loss": 0.4872,
+      "step": 13133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1947605563048983e-05,
+      "loss": 0.447,
+      "step": 13134
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946506295069072e-05,
+      "loss": 0.4885,
+      "step": 13135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194540700264093e-05,
+      "loss": 0.4444,
+      "step": 13136
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194430768577837e-05,
+      "loss": 0.4388,
+      "step": 13137
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943208344495194e-05,
+      "loss": 0.4348,
+      "step": 13138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942108978805213e-05,
+      "loss": 0.4306,
+      "step": 13139
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941009588722236e-05,
+      "loss": 0.4322,
+      "step": 13140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939910174260068e-05,
+      "loss": 0.3892,
+      "step": 13141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193881073543252e-05,
+      "loss": 0.466,
+      "step": 13142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19377112722534e-05,
+      "loss": 0.4258,
+      "step": 13143
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193661178473652e-05,
+      "loss": 0.459,
+      "step": 13144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935512272895688e-05,
+      "loss": 0.4676,
+      "step": 13145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934412736744712e-05,
+      "loss": 0.4866,
+      "step": 13146
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933313176297401e-05,
+      "loss": 0.4804,
+      "step": 13147
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1932213591567571e-05,
+      "loss": 0.4911,
+      "step": 13148
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1931113982569031e-05,
+      "loss": 0.4853,
+      "step": 13149
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.193001434931559e-05,
+      "loss": 0.431,
+      "step": 13150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1928914691821062e-05,
+      "loss": 0.4591,
+      "step": 13151
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1927815010099257e-05,
+      "loss": 0.4692,
+      "step": 13152
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926715304163985e-05,
+      "loss": 0.473,
+      "step": 13153
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192561557402906e-05,
+      "loss": 0.4129,
+      "step": 13154
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19245158197083e-05,
+      "loss": 0.476,
+      "step": 13155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923416041215512e-05,
+      "loss": 0.4632,
+      "step": 13156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192231623856451e-05,
+      "loss": 0.4479,
+      "step": 13157
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192121641176911e-05,
+      "loss": 0.4691,
+      "step": 13158
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920116560843122e-05,
+      "loss": 0.4621,
+      "step": 13159
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919016685800364e-05,
+      "loss": 0.4903,
+      "step": 13160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1917916786654648e-05,
+      "loss": 0.4378,
+      "step": 13161
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916816863419792e-05,
+      "loss": 0.4102,
+      "step": 13162
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915716916109608e-05,
+      "loss": 0.4763,
+      "step": 13163
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191461694473791e-05,
+      "loss": 0.5018,
+      "step": 13164
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1913516949318516e-05,
+      "loss": 0.5168,
+      "step": 13165
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1912416929865245e-05,
+      "loss": 0.4322,
+      "step": 13166
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911316886391908e-05,
+      "loss": 0.4041,
+      "step": 13167
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1910216818912323e-05,
+      "loss": 0.434,
+      "step": 13168
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909116727440311e-05,
+      "loss": 0.4298,
+      "step": 13169
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190801661198968e-05,
+      "loss": 0.5037,
+      "step": 13170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906916472574262e-05,
+      "loss": 0.4878,
+      "step": 13171
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905816309207862e-05,
+      "loss": 0.4605,
+      "step": 13172
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904716121904301e-05,
+      "loss": 0.4579,
+      "step": 13173
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19036159106774e-05,
+      "loss": 0.4618,
+      "step": 13174
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1902515675540977e-05,
+      "loss": 0.4347,
+      "step": 13175
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190141541650885e-05,
+      "loss": 0.4933,
+      "step": 13176
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900315133594839e-05,
+      "loss": 0.509,
+      "step": 13177
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1899214826812764e-05,
+      "loss": 0.4813,
+      "step": 13178
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898114496176443e-05,
+      "loss": 0.4652,
+      "step": 13179
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18970141416997e-05,
+      "loss": 0.4519,
+      "step": 13180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895913763396351e-05,
+      "loss": 0.4968,
+      "step": 13181
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1894813361280222e-05,
+      "loss": 0.4925,
+      "step": 13182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189371293536513e-05,
+      "loss": 0.3919,
+      "step": 13183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892612485664893e-05,
+      "loss": 0.4548,
+      "step": 13184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189151201219334e-05,
+      "loss": 0.4777,
+      "step": 13185
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890411514964291e-05,
+      "loss": 0.374,
+      "step": 13186
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1889310993991567e-05,
+      "loss": 0.4447,
+      "step": 13187
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888210449288992e-05,
+      "loss": 0.4665,
+      "step": 13188
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887109880870389e-05,
+      "loss": 0.4536,
+      "step": 13189
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886009288749577e-05,
+      "loss": 0.5044,
+      "step": 13190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1884908672940382e-05,
+      "loss": 0.4903,
+      "step": 13191
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883808033456632e-05,
+      "loss": 0.5003,
+      "step": 13192
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882707370312147e-05,
+      "loss": 0.4406,
+      "step": 13193
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881606683520753e-05,
+      "loss": 0.4895,
+      "step": 13194
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880505973096267e-05,
+      "loss": 0.468,
+      "step": 13195
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879405239052529e-05,
+      "loss": 0.4384,
+      "step": 13196
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1878304481403353e-05,
+      "loss": 0.4051,
+      "step": 13197
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877203700162566e-05,
+      "loss": 0.4365,
+      "step": 13198
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876102895343997e-05,
+      "loss": 0.4663,
+      "step": 13199
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875002066961469e-05,
+      "loss": 0.4767,
+      "step": 13200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1873901215028812e-05,
+      "loss": 0.4491,
+      "step": 13201
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872800339559851e-05,
+      "loss": 0.4188,
+      "step": 13202
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871699440568411e-05,
+      "loss": 0.4671,
+      "step": 13203
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1870598518068321e-05,
+      "loss": 0.4668,
+      "step": 13204
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186949757207341e-05,
+      "loss": 0.4808,
+      "step": 13205
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868396602597505e-05,
+      "loss": 0.4385,
+      "step": 13206
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867295609654434e-05,
+      "loss": 0.4067,
+      "step": 13207
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1866194593258024e-05,
+      "loss": 0.4227,
+      "step": 13208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865093553422109e-05,
+      "loss": 0.474,
+      "step": 13209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186399249016051e-05,
+      "loss": 0.4953,
+      "step": 13210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862891403487062e-05,
+      "loss": 0.448,
+      "step": 13211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1861790293415596e-05,
+      "loss": 0.3872,
+      "step": 13212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860689159959937e-05,
+      "loss": 0.4662,
+      "step": 13213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185958800313392e-05,
+      "loss": 0.4935,
+      "step": 13214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1858486822951372e-05,
+      "loss": 0.4281,
+      "step": 13215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857385619426124e-05,
+      "loss": 0.4809,
+      "step": 13216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856284392572014e-05,
+      "loss": 0.5429,
+      "step": 13217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1855183142402864e-05,
+      "loss": 0.4832,
+      "step": 13218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185408186893251e-05,
+      "loss": 0.4989,
+      "step": 13219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852980572174786e-05,
+      "loss": 0.4567,
+      "step": 13220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185187925214352e-05,
+      "loss": 0.4317,
+      "step": 13221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1850777908852546e-05,
+      "loss": 0.4883,
+      "step": 13222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18496765423157e-05,
+      "loss": 0.4462,
+      "step": 13223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848575152546812e-05,
+      "loss": 0.4325,
+      "step": 13224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847473739559719e-05,
+      "loss": 0.4285,
+      "step": 13225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1846372303368249e-05,
+      "loss": 0.4998,
+      "step": 13226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184527084398624e-05,
+      "loss": 0.4394,
+      "step": 13227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844169361427526e-05,
+      "loss": 0.4534,
+      "step": 13228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1843067855705944e-05,
+      "loss": 0.4554,
+      "step": 13229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1841966326835323e-05,
+      "loss": 0.4838,
+      "step": 13230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840864774829506e-05,
+      "loss": 0.467,
+      "step": 13231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839763199702319e-05,
+      "loss": 0.4999,
+      "step": 13232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838661601467607e-05,
+      "loss": 0.4571,
+      "step": 13233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837559980139204e-05,
+      "loss": 0.41,
+      "step": 13234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183645833573094e-05,
+      "loss": 0.4492,
+      "step": 13235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1835356668256659e-05,
+      "loss": 0.4357,
+      "step": 13236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834254977730196e-05,
+      "loss": 0.4613,
+      "step": 13237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1833153264165382e-05,
+      "loss": 0.4898,
+      "step": 13238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183205152757607e-05,
+      "loss": 0.4467,
+      "step": 13239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830949767976084e-05,
+      "loss": 0.4698,
+      "step": 13240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1829847985379262e-05,
+      "loss": 0.4593,
+      "step": 13241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1828746179799452e-05,
+      "loss": 0.435,
+      "step": 13242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1827644351250484e-05,
+      "loss": 0.4834,
+      "step": 13243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1826542499746206e-05,
+      "loss": 0.5034,
+      "step": 13244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1825440625300448e-05,
+      "loss": 0.5363,
+      "step": 13245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824338727927054e-05,
+      "loss": 0.4046,
+      "step": 13246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1823236807639861e-05,
+      "loss": 0.492,
+      "step": 13247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1822134864452715e-05,
+      "loss": 0.4792,
+      "step": 13248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821032898379453e-05,
+      "loss": 0.4737,
+      "step": 13249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1819930909433914e-05,
+      "loss": 0.4862,
+      "step": 13250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1818828897629941e-05,
+      "loss": 0.496,
+      "step": 13251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817726862981376e-05,
+      "loss": 0.4377,
+      "step": 13252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181662480550206e-05,
+      "loss": 0.4264,
+      "step": 13253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1815522725205834e-05,
+      "loss": 0.423,
+      "step": 13254
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1814420622106541e-05,
+      "loss": 0.4703,
+      "step": 13255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813318496218025e-05,
+      "loss": 0.4478,
+      "step": 13256
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812216347554126e-05,
+      "loss": 0.4518,
+      "step": 13257
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811114176128685e-05,
+      "loss": 0.4137,
+      "step": 13258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1810011981955552e-05,
+      "loss": 0.4047,
+      "step": 13259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1808909765048567e-05,
+      "loss": 0.4085,
+      "step": 13260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807807525421572e-05,
+      "loss": 0.4789,
+      "step": 13261
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806705263088417e-05,
+      "loss": 0.4233,
+      "step": 13262
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180560297806294e-05,
+      "loss": 0.4138,
+      "step": 13263
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804500670358986e-05,
+      "loss": 0.4578,
+      "step": 13264
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803398339990408e-05,
+      "loss": 0.494,
+      "step": 13265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1802295986971045e-05,
+      "loss": 0.5289,
+      "step": 13266
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801193611314741e-05,
+      "loss": 0.4315,
+      "step": 13267
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1800091213035345e-05,
+      "loss": 0.4239,
+      "step": 13268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798988792146702e-05,
+      "loss": 0.4678,
+      "step": 13269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1797886348662659e-05,
+      "loss": 0.4132,
+      "step": 13270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796783882597068e-05,
+      "loss": 0.4584,
+      "step": 13271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795681393963763e-05,
+      "loss": 0.4652,
+      "step": 13272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1794578882776601e-05,
+      "loss": 0.4322,
+      "step": 13273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179347634904943e-05,
+      "loss": 0.4827,
+      "step": 13274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792373792796095e-05,
+      "loss": 0.4203,
+      "step": 13275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791271214030446e-05,
+      "loss": 0.4391,
+      "step": 13276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790168612766331e-05,
+      "loss": 0.3805,
+      "step": 13277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789065989017592e-05,
+      "loss": 0.4973,
+      "step": 13278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1787963342798088e-05,
+      "loss": 0.4533,
+      "step": 13279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786860674121665e-05,
+      "loss": 0.5414,
+      "step": 13280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785757983002171e-05,
+      "loss": 0.3713,
+      "step": 13281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1784655269453459e-05,
+      "loss": 0.4916,
+      "step": 13282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1783552533489372e-05,
+      "loss": 0.447,
+      "step": 13283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782449775123768e-05,
+      "loss": 0.4613,
+      "step": 13284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781346994370492e-05,
+      "loss": 0.4603,
+      "step": 13285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780244191243404e-05,
+      "loss": 0.3862,
+      "step": 13286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779141365756346e-05,
+      "loss": 0.4821,
+      "step": 13287
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778038517923171e-05,
+      "loss": 0.431,
+      "step": 13288
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1776935647757735e-05,
+      "loss": 0.4618,
+      "step": 13289
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775832755273886e-05,
+      "loss": 0.4913,
+      "step": 13290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177472984048548e-05,
+      "loss": 0.4909,
+      "step": 13291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1773626903406368e-05,
+      "loss": 0.4046,
+      "step": 13292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772523944050403e-05,
+      "loss": 0.4642,
+      "step": 13293
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771420962431437e-05,
+      "loss": 0.4576,
+      "step": 13294
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770317958563322e-05,
+      "loss": 0.4549,
+      "step": 13295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769214932459915e-05,
+      "loss": 0.4357,
+      "step": 13296
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1768111884135074e-05,
+      "loss": 0.5178,
+      "step": 13297
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767008813602645e-05,
+      "loss": 0.4352,
+      "step": 13298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765905720876488e-05,
+      "loss": 0.4706,
+      "step": 13299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1764802605970455e-05,
+      "loss": 0.4823,
+      "step": 13300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1763699468898401e-05,
+      "loss": 0.4324,
+      "step": 13301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762596309674186e-05,
+      "loss": 0.4768,
+      "step": 13302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761493128311663e-05,
+      "loss": 0.4525,
+      "step": 13303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760389924824685e-05,
+      "loss": 0.3782,
+      "step": 13304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175928669922711e-05,
+      "loss": 0.5171,
+      "step": 13305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758183451532797e-05,
+      "loss": 0.4542,
+      "step": 13306
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1757080181755603e-05,
+      "loss": 0.4916,
+      "step": 13307
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1755976889909384e-05,
+      "loss": 0.4245,
+      "step": 13308
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754873576007992e-05,
+      "loss": 0.5477,
+      "step": 13309
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753770240065294e-05,
+      "loss": 0.4275,
+      "step": 13310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1752666882095143e-05,
+      "loss": 0.4981,
+      "step": 13311
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1751563502111395e-05,
+      "loss": 0.4597,
+      "step": 13312
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1750460100127916e-05,
+      "loss": 0.477,
+      "step": 13313
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749356676158559e-05,
+      "loss": 0.4198,
+      "step": 13314
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174825323021718e-05,
+      "loss": 0.485,
+      "step": 13315
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747149762317647e-05,
+      "loss": 0.4638,
+      "step": 13316
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746046272473812e-05,
+      "loss": 0.4538,
+      "step": 13317
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1744942760699541e-05,
+      "loss": 0.4577,
+      "step": 13318
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743839227008688e-05,
+      "loss": 0.4398,
+      "step": 13319
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174273567141512e-05,
+      "loss": 0.4418,
+      "step": 13320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741632093932691e-05,
+      "loss": 0.4513,
+      "step": 13321
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740528494575267e-05,
+      "loss": 0.4374,
+      "step": 13322
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739424873356709e-05,
+      "loss": 0.5098,
+      "step": 13323
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1738321230290874e-05,
+      "loss": 0.5087,
+      "step": 13324
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173721756539163e-05,
+      "loss": 0.466,
+      "step": 13325
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736113878672835e-05,
+      "loss": 0.426,
+      "step": 13326
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173501017014835e-05,
+      "loss": 0.5019,
+      "step": 13327
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733906439832044e-05,
+      "loss": 0.476,
+      "step": 13328
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732802687737774e-05,
+      "loss": 0.4433,
+      "step": 13329
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1731698913879406e-05,
+      "loss": 0.4621,
+      "step": 13330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1730595118270803e-05,
+      "loss": 0.4773,
+      "step": 13331
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729491300925826e-05,
+      "loss": 0.4264,
+      "step": 13332
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1728387461858344e-05,
+      "loss": 0.4509,
+      "step": 13333
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172728360108222e-05,
+      "loss": 0.4595,
+      "step": 13334
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1726179718611313e-05,
+      "loss": 0.4467,
+      "step": 13335
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1725075814459495e-05,
+      "loss": 0.4837,
+      "step": 13336
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723971888640626e-05,
+      "loss": 0.4891,
+      "step": 13337
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1722867941168574e-05,
+      "loss": 0.4216,
+      "step": 13338
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721763972057208e-05,
+      "loss": 0.4158,
+      "step": 13339
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1720659981320386e-05,
+      "loss": 0.4734,
+      "step": 13340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1719555968971979e-05,
+      "loss": 0.5211,
+      "step": 13341
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1718451935025853e-05,
+      "loss": 0.4734,
+      "step": 13342
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1717347879495874e-05,
+      "loss": 0.4594,
+      "step": 13343
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171624380239591e-05,
+      "loss": 0.4515,
+      "step": 13344
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1715139703739829e-05,
+      "loss": 0.4825,
+      "step": 13345
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714035583541491e-05,
+      "loss": 0.457,
+      "step": 13346
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712931441814776e-05,
+      "loss": 0.528,
+      "step": 13347
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1711827278573547e-05,
+      "loss": 0.4774,
+      "step": 13348
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710723093831668e-05,
+      "loss": 0.4827,
+      "step": 13349
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709618887603013e-05,
+      "loss": 0.4719,
+      "step": 13350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708514659901451e-05,
+      "loss": 0.4327,
+      "step": 13351
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707410410740845e-05,
+      "loss": 0.4608,
+      "step": 13352
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170630614013507e-05,
+      "loss": 0.4381,
+      "step": 13353
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705201848097995e-05,
+      "loss": 0.3913,
+      "step": 13354
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170409753464349e-05,
+      "loss": 0.4496,
+      "step": 13355
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702993199785425e-05,
+      "loss": 0.4481,
+      "step": 13356
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1701888843537667e-05,
+      "loss": 0.4962,
+      "step": 13357
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170078446591409e-05,
+      "loss": 0.4726,
+      "step": 13358
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1699680066928567e-05,
+      "loss": 0.4724,
+      "step": 13359
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698575646594967e-05,
+      "loss": 0.4953,
+      "step": 13360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697471204927162e-05,
+      "loss": 0.4108,
+      "step": 13361
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1696366741939022e-05,
+      "loss": 0.4589,
+      "step": 13362
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695262257644421e-05,
+      "loss": 0.4598,
+      "step": 13363
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1694157752057231e-05,
+      "loss": 0.3863,
+      "step": 13364
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1693053225191327e-05,
+      "loss": 0.4725,
+      "step": 13365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691948677060579e-05,
+      "loss": 0.4322,
+      "step": 13366
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169084410767886e-05,
+      "loss": 0.4371,
+      "step": 13367
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1689739517060044e-05,
+      "loss": 0.4902,
+      "step": 13368
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688634905218003e-05,
+      "loss": 0.428,
+      "step": 13369
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1687530272166616e-05,
+      "loss": 0.4843,
+      "step": 13370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1686425617919756e-05,
+      "loss": 0.449,
+      "step": 13371
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685320942491292e-05,
+      "loss": 0.4712,
+      "step": 13372
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684216245895104e-05,
+      "loss": 0.4623,
+      "step": 13373
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1683111528145067e-05,
+      "loss": 0.5091,
+      "step": 13374
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1682006789255052e-05,
+      "loss": 0.5011,
+      "step": 13375
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1680902029238941e-05,
+      "loss": 0.4841,
+      "step": 13376
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1679797248110605e-05,
+      "loss": 0.4423,
+      "step": 13377
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678692445883919e-05,
+      "loss": 0.4559,
+      "step": 13378
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677587622572763e-05,
+      "loss": 0.471,
+      "step": 13379
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676482778191012e-05,
+      "loss": 0.4895,
+      "step": 13380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1675377912752545e-05,
+      "loss": 0.4576,
+      "step": 13381
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674273026271239e-05,
+      "loss": 0.4832,
+      "step": 13382
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673168118760968e-05,
+      "loss": 0.448,
+      "step": 13383
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1672063190235612e-05,
+      "loss": 0.4626,
+      "step": 13384
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1670958240709048e-05,
+      "loss": 0.5038,
+      "step": 13385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1669853270195155e-05,
+      "loss": 0.4512,
+      "step": 13386
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668748278707813e-05,
+      "loss": 0.4723,
+      "step": 13387
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16676432662609e-05,
+      "loss": 0.4166,
+      "step": 13388
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166653823286829e-05,
+      "loss": 0.4493,
+      "step": 13389
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1665433178543868e-05,
+      "loss": 0.5099,
+      "step": 13390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664328103301514e-05,
+      "loss": 0.4132,
+      "step": 13391
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663223007155104e-05,
+      "loss": 0.4283,
+      "step": 13392
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1662117890118521e-05,
+      "loss": 0.3719,
+      "step": 13393
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661012752205646e-05,
+      "loss": 0.5132,
+      "step": 13394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1659907593430354e-05,
+      "loss": 0.4747,
+      "step": 13395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165880241380653e-05,
+      "loss": 0.4697,
+      "step": 13396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657697213348055e-05,
+      "loss": 0.4114,
+      "step": 13397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1656591992068813e-05,
+      "loss": 0.4551,
+      "step": 13398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165548674998268e-05,
+      "loss": 0.4939,
+      "step": 13399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654381487103541e-05,
+      "loss": 0.4911,
+      "step": 13400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653276203445278e-05,
+      "loss": 0.4545,
+      "step": 13401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652170899021776e-05,
+      "loss": 0.4439,
+      "step": 13402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165106557384691e-05,
+      "loss": 0.4703,
+      "step": 13403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1649960227934573e-05,
+      "loss": 0.4894,
+      "step": 13404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164885486129864e-05,
+      "loss": 0.4836,
+      "step": 13405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647749473952996e-05,
+      "loss": 0.4504,
+      "step": 13406
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164664406591153e-05,
+      "loss": 0.477,
+      "step": 13407
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645538637188123e-05,
+      "loss": 0.4934,
+      "step": 13408
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644433187796656e-05,
+      "loss": 0.4552,
+      "step": 13409
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643327717751016e-05,
+      "loss": 0.4465,
+      "step": 13410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164222222706509e-05,
+      "loss": 0.5004,
+      "step": 13411
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641116715752756e-05,
+      "loss": 0.4953,
+      "step": 13412
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164001118382791e-05,
+      "loss": 0.4636,
+      "step": 13413
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638905631304426e-05,
+      "loss": 0.496,
+      "step": 13414
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16378000581962e-05,
+      "loss": 0.4169,
+      "step": 13415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1636694464517114e-05,
+      "loss": 0.4941,
+      "step": 13416
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163558885028105e-05,
+      "loss": 0.463,
+      "step": 13417
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634483215501904e-05,
+      "loss": 0.4681,
+      "step": 13418
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633377560193552e-05,
+      "loss": 0.437,
+      "step": 13419
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632271884369888e-05,
+      "loss": 0.4495,
+      "step": 13420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1631166188044797e-05,
+      "loss": 0.4607,
+      "step": 13421
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1630060471232167e-05,
+      "loss": 0.5182,
+      "step": 13422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162895473394589e-05,
+      "loss": 0.463,
+      "step": 13423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1627848976199848e-05,
+      "loss": 0.441,
+      "step": 13424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162674319800793e-05,
+      "loss": 0.4761,
+      "step": 13425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625637399384026e-05,
+      "loss": 0.4288,
+      "step": 13426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1624531580342029e-05,
+      "loss": 0.4457,
+      "step": 13427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1623425740895822e-05,
+      "loss": 0.4773,
+      "step": 13428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1622319881059297e-05,
+      "loss": 0.4744,
+      "step": 13429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621214000846343e-05,
+      "loss": 0.4903,
+      "step": 13430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162010810027085e-05,
+      "loss": 0.4676,
+      "step": 13431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161900217934671e-05,
+      "loss": 0.4544,
+      "step": 13432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161789623808781e-05,
+      "loss": 0.4891,
+      "step": 13433
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1616790276508045e-05,
+      "loss": 0.4301,
+      "step": 13434
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615684294621302e-05,
+      "loss": 0.4592,
+      "step": 13435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1614578292441476e-05,
+      "loss": 0.4715,
+      "step": 13436
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613472269982453e-05,
+      "loss": 0.5122,
+      "step": 13437
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161236622725813e-05,
+      "loss": 0.5196,
+      "step": 13438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1611260164282397e-05,
+      "loss": 0.4817,
+      "step": 13439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1610154081069144e-05,
+      "loss": 0.4469,
+      "step": 13440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609047977632267e-05,
+      "loss": 0.4078,
+      "step": 13441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607941853985658e-05,
+      "loss": 0.4584,
+      "step": 13442
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606835710143207e-05,
+      "loss": 0.4774,
+      "step": 13443
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605729546118812e-05,
+      "loss": 0.4955,
+      "step": 13444
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1604623361926364e-05,
+      "loss": 0.4938,
+      "step": 13445
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603517157579753e-05,
+      "loss": 0.4291,
+      "step": 13446
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160241093309288e-05,
+      "loss": 0.3838,
+      "step": 13447
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1601304688479638e-05,
+      "loss": 0.4594,
+      "step": 13448
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1600198423753915e-05,
+      "loss": 0.5044,
+      "step": 13449
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1599092138929612e-05,
+      "loss": 0.4334,
+      "step": 13450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597985834020623e-05,
+      "loss": 0.4521,
+      "step": 13451
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159687950904084e-05,
+      "loss": 0.4982,
+      "step": 13452
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1595773164004162e-05,
+      "loss": 0.4777,
+      "step": 13453
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1594666798924482e-05,
+      "loss": 0.4755,
+      "step": 13454
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593560413815703e-05,
+      "loss": 0.4408,
+      "step": 13455
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592454008691714e-05,
+      "loss": 0.4716,
+      "step": 13456
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159134758356641e-05,
+      "loss": 0.4263,
+      "step": 13457
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590241138453693e-05,
+      "loss": 0.4907,
+      "step": 13458
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158913467336746e-05,
+      "loss": 0.4461,
+      "step": 13459
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588028188321606e-05,
+      "loss": 0.4352,
+      "step": 13460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1586921683330029e-05,
+      "loss": 0.4664,
+      "step": 13461
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585815158406626e-05,
+      "loss": 0.4562,
+      "step": 13462
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584708613565291e-05,
+      "loss": 0.4719,
+      "step": 13463
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1583602048819932e-05,
+      "loss": 0.4361,
+      "step": 13464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582495464184444e-05,
+      "loss": 0.4819,
+      "step": 13465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581388859672725e-05,
+      "loss": 0.4269,
+      "step": 13466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158028223529867e-05,
+      "loss": 0.5108,
+      "step": 13467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1579175591076183e-05,
+      "loss": 0.4515,
+      "step": 13468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1578068927019159e-05,
+      "loss": 0.4343,
+      "step": 13469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576962243141506e-05,
+      "loss": 0.4875,
+      "step": 13470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1575855539457118e-05,
+      "loss": 0.4122,
+      "step": 13471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574748815979892e-05,
+      "loss": 0.4656,
+      "step": 13472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573642072723737e-05,
+      "loss": 0.4636,
+      "step": 13473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1572535309702546e-05,
+      "loss": 0.4351,
+      "step": 13474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571428526930227e-05,
+      "loss": 0.406,
+      "step": 13475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1570321724420677e-05,
+      "loss": 0.4256,
+      "step": 13476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1569214902187795e-05,
+      "loss": 0.4679,
+      "step": 13477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1568108060245488e-05,
+      "loss": 0.4605,
+      "step": 13478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567001198607656e-05,
+      "loss": 0.4822,
+      "step": 13479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1565894317288202e-05,
+      "loss": 0.4519,
+      "step": 13480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156478741630103e-05,
+      "loss": 0.4858,
+      "step": 13481
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563680495660036e-05,
+      "loss": 0.4972,
+      "step": 13482
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1562573555379127e-05,
+      "loss": 0.4878,
+      "step": 13483
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156146659547221e-05,
+      "loss": 0.4381,
+      "step": 13484
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560359615953182e-05,
+      "loss": 0.4881,
+      "step": 13485
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1559252616835952e-05,
+      "loss": 0.4321,
+      "step": 13486
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1558145598134423e-05,
+      "loss": 0.412,
+      "step": 13487
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1557038559862495e-05,
+      "loss": 0.4883,
+      "step": 13488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555931502034077e-05,
+      "loss": 0.4269,
+      "step": 13489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554824424663072e-05,
+      "loss": 0.4587,
+      "step": 13490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553717327763386e-05,
+      "loss": 0.4829,
+      "step": 13491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552610211348921e-05,
+      "loss": 0.4767,
+      "step": 13492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1551503075433587e-05,
+      "loss": 0.4772,
+      "step": 13493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550395920031283e-05,
+      "loss": 0.4677,
+      "step": 13494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549288745155924e-05,
+      "loss": 0.519,
+      "step": 13495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548181550821411e-05,
+      "loss": 0.5053,
+      "step": 13496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1547074337041652e-05,
+      "loss": 0.4654,
+      "step": 13497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1545967103830551e-05,
+      "loss": 0.491,
+      "step": 13498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544859851202018e-05,
+      "loss": 0.4919,
+      "step": 13499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543752579169955e-05,
+      "loss": 0.4387,
+      "step": 13500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1542645287748278e-05,
+      "loss": 0.4788,
+      "step": 13501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1541537976950887e-05,
+      "loss": 0.5018,
+      "step": 13502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1540430646791693e-05,
+      "loss": 0.4608,
+      "step": 13503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539323297284603e-05,
+      "loss": 0.394,
+      "step": 13504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153821592844353e-05,
+      "loss": 0.4725,
+      "step": 13505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537108540282373e-05,
+      "loss": 0.4645,
+      "step": 13506
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536001132815051e-05,
+      "loss": 0.4757,
+      "step": 13507
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534893706055468e-05,
+      "loss": 0.4988,
+      "step": 13508
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533786260017532e-05,
+      "loss": 0.4804,
+      "step": 13509
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532678794715158e-05,
+      "loss": 0.4476,
+      "step": 13510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1531571310162248e-05,
+      "loss": 0.4756,
+      "step": 13511
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530463806372721e-05,
+      "loss": 0.4725,
+      "step": 13512
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529356283360483e-05,
+      "loss": 0.4395,
+      "step": 13513
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152824874113944e-05,
+      "loss": 0.4656,
+      "step": 13514
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152714117972351e-05,
+      "loss": 0.4261,
+      "step": 13515
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526033599126603e-05,
+      "loss": 0.4651,
+      "step": 13516
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1524925999362624e-05,
+      "loss": 0.4784,
+      "step": 13517
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523818380445494e-05,
+      "loss": 0.4195,
+      "step": 13518
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522710742389118e-05,
+      "loss": 0.4998,
+      "step": 13519
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521603085207406e-05,
+      "loss": 0.4571,
+      "step": 13520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1520495408914278e-05,
+      "loss": 0.4459,
+      "step": 13521
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519387713523643e-05,
+      "loss": 0.475,
+      "step": 13522
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151827999904941e-05,
+      "loss": 0.4509,
+      "step": 13523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1517172265505498e-05,
+      "loss": 0.4274,
+      "step": 13524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1516064512905815e-05,
+      "loss": 0.4511,
+      "step": 13525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151495674126428e-05,
+      "loss": 0.426,
+      "step": 13526
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513848950594801e-05,
+      "loss": 0.4572,
+      "step": 13527
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512741140911296e-05,
+      "loss": 0.4116,
+      "step": 13528
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511633312227676e-05,
+      "loss": 0.4527,
+      "step": 13529
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151052546455786e-05,
+      "loss": 0.4065,
+      "step": 13530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509417597915755e-05,
+      "loss": 0.4834,
+      "step": 13531
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1508309712315282e-05,
+      "loss": 0.457,
+      "step": 13532
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1507201807770356e-05,
+      "loss": 0.4277,
+      "step": 13533
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506093884294891e-05,
+      "loss": 0.416,
+      "step": 13534
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1504985941902802e-05,
+      "loss": 0.4361,
+      "step": 13535
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503877980608006e-05,
+      "loss": 0.4495,
+      "step": 13536
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1502770000424415e-05,
+      "loss": 0.4291,
+      "step": 13537
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150166200136595e-05,
+      "loss": 0.4661,
+      "step": 13538
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1500553983446527e-05,
+      "loss": 0.4683,
+      "step": 13539
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1499445946680063e-05,
+      "loss": 0.4612,
+      "step": 13540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498337891080474e-05,
+      "loss": 0.4859,
+      "step": 13541
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1497229816661673e-05,
+      "loss": 0.4763,
+      "step": 13542
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496121723437585e-05,
+      "loss": 0.4578,
+      "step": 13543
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1495013611422125e-05,
+      "loss": 0.4939,
+      "step": 13544
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149390548062921e-05,
+      "loss": 0.4908,
+      "step": 13545
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1492797331072758e-05,
+      "loss": 0.4183,
+      "step": 13546
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491689162766688e-05,
+      "loss": 0.4719,
+      "step": 13547
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490580975724916e-05,
+      "loss": 0.4783,
+      "step": 13548
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489472769961368e-05,
+      "loss": 0.4991,
+      "step": 13549
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488364545489956e-05,
+      "loss": 0.4759,
+      "step": 13550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1487256302324601e-05,
+      "loss": 0.4459,
+      "step": 13551
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1486148040479226e-05,
+      "loss": 0.4502,
+      "step": 13552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148503975996775e-05,
+      "loss": 0.4874,
+      "step": 13553
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1483931460804085e-05,
+      "loss": 0.4264,
+      "step": 13554
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482823143002164e-05,
+      "loss": 0.5032,
+      "step": 13555
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481714806575898e-05,
+      "loss": 0.4814,
+      "step": 13556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1480606451539211e-05,
+      "loss": 0.4715,
+      "step": 13557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479498077906023e-05,
+      "loss": 0.4789,
+      "step": 13558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147838968569026e-05,
+      "loss": 0.4926,
+      "step": 13559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1477281274905838e-05,
+      "loss": 0.4309,
+      "step": 13560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147617284556668e-05,
+      "loss": 0.4472,
+      "step": 13561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475064397686708e-05,
+      "loss": 0.4847,
+      "step": 13562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473955931279846e-05,
+      "loss": 0.477,
+      "step": 13563
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1472847446360014e-05,
+      "loss": 0.4953,
+      "step": 13564
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471738942941136e-05,
+      "loss": 0.4531,
+      "step": 13565
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470630421037134e-05,
+      "loss": 0.4754,
+      "step": 13566
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146952188066193e-05,
+      "loss": 0.4308,
+      "step": 13567
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146841332182945e-05,
+      "loss": 0.4818,
+      "step": 13568
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467304744553618e-05,
+      "loss": 0.4428,
+      "step": 13569
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1466196148848356e-05,
+      "loss": 0.4676,
+      "step": 13570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465087534727587e-05,
+      "loss": 0.4721,
+      "step": 13571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146397890220524e-05,
+      "loss": 0.4674,
+      "step": 13572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146287025129523e-05,
+      "loss": 0.4868,
+      "step": 13573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146176158201149e-05,
+      "loss": 0.4633,
+      "step": 13574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460652894367946e-05,
+      "loss": 0.4235,
+      "step": 13575
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459544188378516e-05,
+      "loss": 0.5021,
+      "step": 13576
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458435464057132e-05,
+      "loss": 0.4358,
+      "step": 13577
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457326721417715e-05,
+      "loss": 0.4771,
+      "step": 13578
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456217960474192e-05,
+      "loss": 0.467,
+      "step": 13579
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455109181240491e-05,
+      "loss": 0.4898,
+      "step": 13580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1454000383730537e-05,
+      "loss": 0.4272,
+      "step": 13581
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452891567958258e-05,
+      "loss": 0.4989,
+      "step": 13582
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451782733937578e-05,
+      "loss": 0.4594,
+      "step": 13583
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450673881682426e-05,
+      "loss": 0.45,
+      "step": 13584
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449565011206728e-05,
+      "loss": 0.4227,
+      "step": 13585
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448456122524414e-05,
+      "loss": 0.4795,
+      "step": 13586
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1447347215649411e-05,
+      "loss": 0.4653,
+      "step": 13587
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446238290595642e-05,
+      "loss": 0.4205,
+      "step": 13588
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1445129347377042e-05,
+      "loss": 0.3834,
+      "step": 13589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444020386007534e-05,
+      "loss": 0.5012,
+      "step": 13590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442911406501049e-05,
+      "loss": 0.4294,
+      "step": 13591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441802408871521e-05,
+      "loss": 0.4331,
+      "step": 13592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440693393132871e-05,
+      "loss": 0.4594,
+      "step": 13593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439584359299029e-05,
+      "loss": 0.4575,
+      "step": 13594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1438475307383929e-05,
+      "loss": 0.5015,
+      "step": 13595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1437366237401499e-05,
+      "loss": 0.4851,
+      "step": 13596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436257149365668e-05,
+      "loss": 0.4157,
+      "step": 13597
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435148043290368e-05,
+      "loss": 0.4276,
+      "step": 13598
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434038919189522e-05,
+      "loss": 0.4998,
+      "step": 13599
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432929777077073e-05,
+      "loss": 0.4187,
+      "step": 13600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431820616966945e-05,
+      "loss": 0.4455,
+      "step": 13601
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143071143887307e-05,
+      "loss": 0.4435,
+      "step": 13602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429602242809379e-05,
+      "loss": 0.4064,
+      "step": 13603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1428493028789803e-05,
+      "loss": 0.4696,
+      "step": 13604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427383796828273e-05,
+      "loss": 0.4419,
+      "step": 13605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426274546938725e-05,
+      "loss": 0.4497,
+      "step": 13606
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142516527913509e-05,
+      "loss": 0.4887,
+      "step": 13607
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424055993431296e-05,
+      "loss": 0.4602,
+      "step": 13608
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142294668984128e-05,
+      "loss": 0.4276,
+      "step": 13609
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1421837368378973e-05,
+      "loss": 0.4623,
+      "step": 13610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142072802905831e-05,
+      "loss": 0.4593,
+      "step": 13611
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419618671893225e-05,
+      "loss": 0.4153,
+      "step": 13612
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141850929689765e-05,
+      "loss": 0.495,
+      "step": 13613
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417399904085516e-05,
+      "loss": 0.4083,
+      "step": 13614
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141629049347076e-05,
+      "loss": 0.4239,
+      "step": 13615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1415181065067318e-05,
+      "loss": 0.4486,
+      "step": 13616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1414071618889123e-05,
+      "loss": 0.4573,
+      "step": 13617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1412962154950109e-05,
+      "loss": 0.4829,
+      "step": 13618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411852673264208e-05,
+      "loss": 0.4933,
+      "step": 13619
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410743173845362e-05,
+      "loss": 0.5132,
+      "step": 13620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409633656707502e-05,
+      "loss": 0.4669,
+      "step": 13621
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1408524121864561e-05,
+      "loss": 0.409,
+      "step": 13622
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1407414569330484e-05,
+      "loss": 0.443,
+      "step": 13623
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406304999119197e-05,
+      "loss": 0.4572,
+      "step": 13624
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140519541124464e-05,
+      "loss": 0.4757,
+      "step": 13625
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404085805720752e-05,
+      "loss": 0.4109,
+      "step": 13626
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1402976182561467e-05,
+      "loss": 0.465,
+      "step": 13627
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1401866541780724e-05,
+      "loss": 0.4706,
+      "step": 13628
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1400756883392457e-05,
+      "loss": 0.4045,
+      "step": 13629
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1399647207410607e-05,
+      "loss": 0.4576,
+      "step": 13630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1398537513849106e-05,
+      "loss": 0.5078,
+      "step": 13631
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1397427802721897e-05,
+      "loss": 0.4748,
+      "step": 13632
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396318074042918e-05,
+      "loss": 0.508,
+      "step": 13633
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1395208327826105e-05,
+      "loss": 0.4665,
+      "step": 13634
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1394098564085399e-05,
+      "loss": 0.4528,
+      "step": 13635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392988782834733e-05,
+      "loss": 0.4523,
+      "step": 13636
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1391878984088052e-05,
+      "loss": 0.5137,
+      "step": 13637
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390769167859292e-05,
+      "loss": 0.4948,
+      "step": 13638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389659334162395e-05,
+      "loss": 0.436,
+      "step": 13639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1388549483011299e-05,
+      "loss": 0.4804,
+      "step": 13640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1387439614419943e-05,
+      "loss": 0.4121,
+      "step": 13641
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386329728402269e-05,
+      "loss": 0.4222,
+      "step": 13642
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385219824972214e-05,
+      "loss": 0.4209,
+      "step": 13643
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384109904143721e-05,
+      "loss": 0.4281,
+      "step": 13644
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138299996593073e-05,
+      "loss": 0.4492,
+      "step": 13645
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381890010347183e-05,
+      "loss": 0.5045,
+      "step": 13646
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138078003740702e-05,
+      "loss": 0.436,
+      "step": 13647
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137967004712418e-05,
+      "loss": 0.4882,
+      "step": 13648
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137856003951261e-05,
+      "loss": 0.4629,
+      "step": 13649
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1377450014586249e-05,
+      "loss": 0.4149,
+      "step": 13650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1376339972359034e-05,
+      "loss": 0.445,
+      "step": 13651
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1375229912844915e-05,
+      "loss": 0.3721,
+      "step": 13652
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374119836057832e-05,
+      "loss": 0.3891,
+      "step": 13653
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1373009742011725e-05,
+      "loss": 0.4656,
+      "step": 13654
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1371899630720542e-05,
+      "loss": 0.4183,
+      "step": 13655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1370789502198219e-05,
+      "loss": 0.4415,
+      "step": 13656
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1369679356458705e-05,
+      "loss": 0.4393,
+      "step": 13657
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136856919351594e-05,
+      "loss": 0.5026,
+      "step": 13658
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136745901338387e-05,
+      "loss": 0.4762,
+      "step": 13659
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366348816076442e-05,
+      "loss": 0.4283,
+      "step": 13660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365238601607593e-05,
+      "loss": 0.4305,
+      "step": 13661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136412836999127e-05,
+      "loss": 0.4788,
+      "step": 13662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363018121241418e-05,
+      "loss": 0.3885,
+      "step": 13663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1361907855371984e-05,
+      "loss": 0.4237,
+      "step": 13664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360797572396912e-05,
+      "loss": 0.5138,
+      "step": 13665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359687272330144e-05,
+      "loss": 0.4927,
+      "step": 13666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135857695518563e-05,
+      "loss": 0.4253,
+      "step": 13667
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357466620977309e-05,
+      "loss": 0.4299,
+      "step": 13668
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356356269719133e-05,
+      "loss": 0.5198,
+      "step": 13669
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355245901425047e-05,
+      "loss": 0.4373,
+      "step": 13670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354135516108997e-05,
+      "loss": 0.4865,
+      "step": 13671
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135302511378493e-05,
+      "loss": 0.4582,
+      "step": 13672
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351914694466788e-05,
+      "loss": 0.4807,
+      "step": 13673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350804258168525e-05,
+      "loss": 0.4556,
+      "step": 13674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349693804904085e-05,
+      "loss": 0.4411,
+      "step": 13675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1348583334687415e-05,
+      "loss": 0.459,
+      "step": 13676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134747284753246e-05,
+      "loss": 0.4823,
+      "step": 13677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346362343453174e-05,
+      "loss": 0.4608,
+      "step": 13678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1345251822463499e-05,
+      "loss": 0.4654,
+      "step": 13679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344141284577387e-05,
+      "loss": 0.4561,
+      "step": 13680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1343030729808786e-05,
+      "loss": 0.4304,
+      "step": 13681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1341920158171641e-05,
+      "loss": 0.3964,
+      "step": 13682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1340809569679908e-05,
+      "loss": 0.4167,
+      "step": 13683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133969896434753e-05,
+      "loss": 0.5051,
+      "step": 13684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338588342188455e-05,
+      "loss": 0.4322,
+      "step": 13685
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133747770321664e-05,
+      "loss": 0.439,
+      "step": 13686
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336367047446029e-05,
+      "loss": 0.4313,
+      "step": 13687
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1335256374890571e-05,
+      "loss": 0.4641,
+      "step": 13688
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1334145685564217e-05,
+      "loss": 0.4708,
+      "step": 13689
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333034979480919e-05,
+      "loss": 0.4332,
+      "step": 13690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133192425665463e-05,
+      "loss": 0.4657,
+      "step": 13691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13308135170993e-05,
+      "loss": 0.3914,
+      "step": 13692
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132970276082887e-05,
+      "loss": 0.4163,
+      "step": 13693
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328591987857305e-05,
+      "loss": 0.416,
+      "step": 13694
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327481198198548e-05,
+      "loss": 0.4411,
+      "step": 13695
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132637039186655e-05,
+      "loss": 0.483,
+      "step": 13696
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325259568875274e-05,
+      "loss": 0.5389,
+      "step": 13697
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324148729238656e-05,
+      "loss": 0.4419,
+      "step": 13698
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323037872970657e-05,
+      "loss": 0.4367,
+      "step": 13699
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1321927000085228e-05,
+      "loss": 0.4337,
+      "step": 13700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320816110596325e-05,
+      "loss": 0.4719,
+      "step": 13701
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319705204517894e-05,
+      "loss": 0.4731,
+      "step": 13702
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1318594281863895e-05,
+      "loss": 0.4642,
+      "step": 13703
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1317483342648276e-05,
+      "loss": 0.4577,
+      "step": 13704
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316372386884991e-05,
+      "loss": 0.5082,
+      "step": 13705
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315261414587997e-05,
+      "loss": 0.4168,
+      "step": 13706
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314150425771246e-05,
+      "loss": 0.4769,
+      "step": 13707
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313039420448693e-05,
+      "loss": 0.4666,
+      "step": 13708
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131192839863429e-05,
+      "loss": 0.4845,
+      "step": 13709
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1310817360341992e-05,
+      "loss": 0.4567,
+      "step": 13710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309706305585754e-05,
+      "loss": 0.4862,
+      "step": 13711
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1308595234379536e-05,
+      "loss": 0.4452,
+      "step": 13712
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307484146737285e-05,
+      "loss": 0.4876,
+      "step": 13713
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130637304267296e-05,
+      "loss": 0.465,
+      "step": 13714
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.471,
+      "step": 13715
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130415078533391e-05,
+      "loss": 0.469,
+      "step": 13716
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303039632087099e-05,
+      "loss": 0.4784,
+      "step": 13717
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301928462474034e-05,
+      "loss": 0.4674,
+      "step": 13718
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300817276508678e-05,
+      "loss": 0.4576,
+      "step": 13719
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299706074204981e-05,
+      "loss": 0.4884,
+      "step": 13720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1298594855576905e-05,
+      "loss": 0.4773,
+      "step": 13721
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1297483620638401e-05,
+      "loss": 0.4433,
+      "step": 13722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1296372369403434e-05,
+      "loss": 0.444,
+      "step": 13723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129526110188596e-05,
+      "loss": 0.4848,
+      "step": 13724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294149818099928e-05,
+      "loss": 0.5028,
+      "step": 13725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1293038518059305e-05,
+      "loss": 0.4684,
+      "step": 13726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1291927201778044e-05,
+      "loss": 0.4461,
+      "step": 13727
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1290815869270105e-05,
+      "loss": 0.4608,
+      "step": 13728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128970452054945e-05,
+      "loss": 0.4717,
+      "step": 13729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128859315563003e-05,
+      "loss": 0.4119,
+      "step": 13730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128748177452581e-05,
+      "loss": 0.4299,
+      "step": 13731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286370377250747e-05,
+      "loss": 0.4701,
+      "step": 13732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1285258963818794e-05,
+      "loss": 0.4493,
+      "step": 13733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284147534243925e-05,
+      "loss": 0.4456,
+      "step": 13734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283036088540088e-05,
+      "loss": 0.4339,
+      "step": 13735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281924626721243e-05,
+      "loss": 0.4621,
+      "step": 13736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280813148801357e-05,
+      "loss": 0.4833,
+      "step": 13737
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1279701654794384e-05,
+      "loss": 0.4595,
+      "step": 13738
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278590144714288e-05,
+      "loss": 0.4415,
+      "step": 13739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277478618575025e-05,
+      "loss": 0.4464,
+      "step": 13740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1276367076390565e-05,
+      "loss": 0.4533,
+      "step": 13741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1275255518174857e-05,
+      "loss": 0.4074,
+      "step": 13742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127414394394187e-05,
+      "loss": 0.4598,
+      "step": 13743
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1273032353705564e-05,
+      "loss": 0.4386,
+      "step": 13744
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12719207474799e-05,
+      "loss": 0.4726,
+      "step": 13745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1270809125278843e-05,
+      "loss": 0.5191,
+      "step": 13746
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269697487116347e-05,
+      "loss": 0.5048,
+      "step": 13747
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268585833006383e-05,
+      "loss": 0.5266,
+      "step": 13748
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126747416296291e-05,
+      "loss": 0.5081,
+      "step": 13749
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126636247699989e-05,
+      "loss": 0.4528,
+      "step": 13750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265250775131287e-05,
+      "loss": 0.4789,
+      "step": 13751
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1264139057371062e-05,
+      "loss": 0.5102,
+      "step": 13752
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263027323733178e-05,
+      "loss": 0.4559,
+      "step": 13753
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261915574231603e-05,
+      "loss": 0.4746,
+      "step": 13754
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260803808880299e-05,
+      "loss": 0.4597,
+      "step": 13755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259692027693225e-05,
+      "loss": 0.4622,
+      "step": 13756
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125858023068435e-05,
+      "loss": 0.4238,
+      "step": 13757
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1257468417867639e-05,
+      "loss": 0.4627,
+      "step": 13758
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125635658925705e-05,
+      "loss": 0.4835,
+      "step": 13759
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255244744866556e-05,
+      "loss": 0.4798,
+      "step": 13760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1254132884710113e-05,
+      "loss": 0.5092,
+      "step": 13761
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1253021008801695e-05,
+      "loss": 0.467,
+      "step": 13762
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125190911715526e-05,
+      "loss": 0.4584,
+      "step": 13763
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250797209784776e-05,
+      "loss": 0.4753,
+      "step": 13764
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1249685286704212e-05,
+      "loss": 0.467,
+      "step": 13765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1248573347927532e-05,
+      "loss": 0.4491,
+      "step": 13766
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1247461393468697e-05,
+      "loss": 0.4633,
+      "step": 13767
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1246349423341678e-05,
+      "loss": 0.408,
+      "step": 13768
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124523743756044e-05,
+      "loss": 0.499,
+      "step": 13769
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244125436138953e-05,
+      "loss": 0.5093,
+      "step": 13770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243013419091178e-05,
+      "loss": 0.4785,
+      "step": 13771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241901386431087e-05,
+      "loss": 0.5155,
+      "step": 13772
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240789338172642e-05,
+      "loss": 0.4669,
+      "step": 13773
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1239677274329815e-05,
+      "loss": 0.4818,
+      "step": 13774
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238565194916571e-05,
+      "loss": 0.4614,
+      "step": 13775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123745309994688e-05,
+      "loss": 0.4728,
+      "step": 13776
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123634098943471e-05,
+      "loss": 0.4876,
+      "step": 13777
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1235228863394028e-05,
+      "loss": 0.4464,
+      "step": 13778
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234116721838798e-05,
+      "loss": 0.4395,
+      "step": 13779
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233004564782997e-05,
+      "loss": 0.4993,
+      "step": 13780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231892392240586e-05,
+      "loss": 0.4221,
+      "step": 13781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230780204225542e-05,
+      "loss": 0.3974,
+      "step": 13782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1229668000751827e-05,
+      "loss": 0.4425,
+      "step": 13783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1228555781833413e-05,
+      "loss": 0.493,
+      "step": 13784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122744354748427e-05,
+      "loss": 0.4386,
+      "step": 13785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226331297718367e-05,
+      "loss": 0.4038,
+      "step": 13786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1225219032549675e-05,
+      "loss": 0.4558,
+      "step": 13787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224106751992164e-05,
+      "loss": 0.4146,
+      "step": 13788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1222994456059804e-05,
+      "loss": 0.4438,
+      "step": 13789
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221882144766561e-05,
+      "loss": 0.4841,
+      "step": 13790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1220769818126414e-05,
+      "loss": 0.4819,
+      "step": 13791
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121965747615333e-05,
+      "loss": 0.4526,
+      "step": 13792
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1218545118861275e-05,
+      "loss": 0.4225,
+      "step": 13793
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217432746264229e-05,
+      "loss": 0.4741,
+      "step": 13794
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216320358376158e-05,
+      "loss": 0.4857,
+      "step": 13795
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1215207955211037e-05,
+      "loss": 0.4485,
+      "step": 13796
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1214095536782837e-05,
+      "loss": 0.4604,
+      "step": 13797
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212983103105524e-05,
+      "loss": 0.3895,
+      "step": 13798
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211870654193078e-05,
+      "loss": 0.4867,
+      "step": 13799
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121075819005947e-05,
+      "loss": 0.4658,
+      "step": 13800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1209645710718668e-05,
+      "loss": 0.4287,
+      "step": 13801
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208533216184655e-05,
+      "loss": 0.4039,
+      "step": 13802
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.120742070647139e-05,
+      "loss": 0.4587,
+      "step": 13803
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206308181592856e-05,
+      "loss": 0.4555,
+      "step": 13804
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205195641563022e-05,
+      "loss": 0.486,
+      "step": 13805
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204083086395865e-05,
+      "loss": 0.4409,
+      "step": 13806
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1202970516105358e-05,
+      "loss": 0.4879,
+      "step": 13807
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201857930705473e-05,
+      "loss": 0.4914,
+      "step": 13808
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200745330210186e-05,
+      "loss": 0.4472,
+      "step": 13809
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199632714633466e-05,
+      "loss": 0.468,
+      "step": 13810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1198520083989294e-05,
+      "loss": 0.4646,
+      "step": 13811
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197407438291645e-05,
+      "loss": 0.4514,
+      "step": 13812
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119629477755449e-05,
+      "loss": 0.4417,
+      "step": 13813
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1195182101791808e-05,
+      "loss": 0.4826,
+      "step": 13814
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119406941101757e-05,
+      "loss": 0.4707,
+      "step": 13815
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119295670524575e-05,
+      "loss": 0.4287,
+      "step": 13816
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1191843984490331e-05,
+      "loss": 0.4611,
+      "step": 13817
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1190731248765283e-05,
+      "loss": 0.453,
+      "step": 13818
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1189618498084587e-05,
+      "loss": 0.4704,
+      "step": 13819
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188505732462214e-05,
+      "loss": 0.4313,
+      "step": 13820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1187392951912139e-05,
+      "loss": 0.4732,
+      "step": 13821
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186280156448346e-05,
+      "loss": 0.4227,
+      "step": 13822
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1185167346084807e-05,
+      "loss": 0.4802,
+      "step": 13823
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1184054520835498e-05,
+      "loss": 0.439,
+      "step": 13824
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11829416807144e-05,
+      "loss": 0.4753,
+      "step": 13825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118182882573549e-05,
+      "loss": 0.5042,
+      "step": 13826
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118071595591274e-05,
+      "loss": 0.4782,
+      "step": 13827
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179603071260132e-05,
+      "loss": 0.4865,
+      "step": 13828
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178490171791647e-05,
+      "loss": 0.4831,
+      "step": 13829
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1177377257521256e-05,
+      "loss": 0.4217,
+      "step": 13830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1176264328462943e-05,
+      "loss": 0.4646,
+      "step": 13831
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175151384630682e-05,
+      "loss": 0.4738,
+      "step": 13832
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174038426038452e-05,
+      "loss": 0.5184,
+      "step": 13833
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172925452700238e-05,
+      "loss": 0.4193,
+      "step": 13834
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1171812464630011e-05,
+      "loss": 0.3979,
+      "step": 13835
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170699461841757e-05,
+      "loss": 0.4509,
+      "step": 13836
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169586444349451e-05,
+      "loss": 0.4937,
+      "step": 13837
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168473412167072e-05,
+      "loss": 0.4838,
+      "step": 13838
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167360365308603e-05,
+      "loss": 0.4589,
+      "step": 13839
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1166247303788024e-05,
+      "loss": 0.4281,
+      "step": 13840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116513422761931e-05,
+      "loss": 0.5314,
+      "step": 13841
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164021136816447e-05,
+      "loss": 0.455,
+      "step": 13842
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162908031393414e-05,
+      "loss": 0.5033,
+      "step": 13843
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116179491136419e-05,
+      "loss": 0.4815,
+      "step": 13844
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116068177674276e-05,
+      "loss": 0.4601,
+      "step": 13845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1159568627543099e-05,
+      "loss": 0.4679,
+      "step": 13846
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1158455463779187e-05,
+      "loss": 0.5093,
+      "step": 13847
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157342285465013e-05,
+      "loss": 0.4666,
+      "step": 13848
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1156229092614556e-05,
+      "loss": 0.4367,
+      "step": 13849
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11551158852418e-05,
+      "loss": 0.4597,
+      "step": 13850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115400266336072e-05,
+      "loss": 0.4677,
+      "step": 13851
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152889426985298e-05,
+      "loss": 0.455,
+      "step": 13852
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1151776176129525e-05,
+      "loss": 0.4373,
+      "step": 13853
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115066291080738e-05,
+      "loss": 0.42,
+      "step": 13854
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149549631032842e-05,
+      "loss": 0.5048,
+      "step": 13855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148436336819897e-05,
+      "loss": 0.4372,
+      "step": 13856
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147323028182526e-05,
+      "loss": 0.4688,
+      "step": 13857
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1146209705134715e-05,
+      "loss": 0.4669,
+      "step": 13858
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1145096367690445e-05,
+      "loss": 0.4432,
+      "step": 13859
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11439830158637e-05,
+      "loss": 0.4612,
+      "step": 13860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142869649668467e-05,
+      "loss": 0.4708,
+      "step": 13861
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141756269118724e-05,
+      "loss": 0.4547,
+      "step": 13862
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114064287422846e-05,
+      "loss": 0.4437,
+      "step": 13863
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1139529465011655e-05,
+      "loss": 0.3843,
+      "step": 13864
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138416041482299e-05,
+      "loss": 0.4567,
+      "step": 13865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1137302603654376e-05,
+      "loss": 0.3661,
+      "step": 13866
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136189151541863e-05,
+      "loss": 0.4756,
+      "step": 13867
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1135075685158754e-05,
+      "loss": 0.4709,
+      "step": 13868
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113396220451903e-05,
+      "loss": 0.4555,
+      "step": 13869
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1132848709636677e-05,
+      "loss": 0.4419,
+      "step": 13870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131735200525682e-05,
+      "loss": 0.4555,
+      "step": 13871
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1130621677200026e-05,
+      "loss": 0.4195,
+      "step": 13872
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129508139673704e-05,
+      "loss": 0.446,
+      "step": 13873
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128394587960694e-05,
+      "loss": 0.4564,
+      "step": 13874
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112728102207498e-05,
+      "loss": 0.4131,
+      "step": 13875
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1126167442030562e-05,
+      "loss": 0.44,
+      "step": 13876
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125053847841412e-05,
+      "loss": 0.4943,
+      "step": 13877
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123940239521522e-05,
+      "loss": 0.398,
+      "step": 13878
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1122826617084884e-05,
+      "loss": 0.3679,
+      "step": 13879
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121712980545479e-05,
+      "loss": 0.4306,
+      "step": 13880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1120599329917294e-05,
+      "loss": 0.4675,
+      "step": 13881
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119485665214323e-05,
+      "loss": 0.4683,
+      "step": 13882
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1118371986450544e-05,
+      "loss": 0.4415,
+      "step": 13883
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117258293639953e-05,
+      "loss": 0.4941,
+      "step": 13884
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1116144586796535e-05,
+      "loss": 0.4714,
+      "step": 13885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111503086593428e-05,
+      "loss": 0.4438,
+      "step": 13886
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113917131067173e-05,
+      "loss": 0.4987,
+      "step": 13887
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112803382209206e-05,
+      "loss": 0.5143,
+      "step": 13888
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111689619374361e-05,
+      "loss": 0.4557,
+      "step": 13889
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1110575842576639e-05,
+      "loss": 0.4613,
+      "step": 13890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109462051830018e-05,
+      "loss": 0.4752,
+      "step": 13891
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1108348247148493e-05,
+      "loss": 0.4346,
+      "step": 13892
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107234428546053e-05,
+      "loss": 0.4568,
+      "step": 13893
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1106120596036685e-05,
+      "loss": 0.3842,
+      "step": 13894
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110500674963438e-05,
+      "loss": 0.416,
+      "step": 13895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110389288935313e-05,
+      "loss": 0.4521,
+      "step": 13896
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102779015206924e-05,
+      "loss": 0.4443,
+      "step": 13897
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110166512720975e-05,
+      "loss": 0.4314,
+      "step": 13898
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11005512253756e-05,
+      "loss": 0.5035,
+      "step": 13899
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099437309718467e-05,
+      "loss": 0.4517,
+      "step": 13900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098323380252336e-05,
+      "loss": 0.4536,
+      "step": 13901
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1097209436991205e-05,
+      "loss": 0.465,
+      "step": 13902
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096095479949063e-05,
+      "loss": 0.4869,
+      "step": 13903
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094981509139898e-05,
+      "loss": 0.5099,
+      "step": 13904
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1093867524577705e-05,
+      "loss": 0.467,
+      "step": 13905
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092753526276474e-05,
+      "loss": 0.4362,
+      "step": 13906
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1091639514250198e-05,
+      "loss": 0.4931,
+      "step": 13907
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109052548851287e-05,
+      "loss": 0.4968,
+      "step": 13908
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089411449078475e-05,
+      "loss": 0.4285,
+      "step": 13909
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1088297395961017e-05,
+      "loss": 0.4295,
+      "step": 13910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108718332917448e-05,
+      "loss": 0.4751,
+      "step": 13911
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086069248732862e-05,
+      "loss": 0.4303,
+      "step": 13912
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084955154650152e-05,
+      "loss": 0.4479,
+      "step": 13913
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1083841046940344e-05,
+      "loss": 0.4351,
+      "step": 13914
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1082726925617431e-05,
+      "loss": 0.4154,
+      "step": 13915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1081612790695409e-05,
+      "loss": 0.4778,
+      "step": 13916
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108049864218827e-05,
+      "loss": 0.4446,
+      "step": 13917
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079384480110006e-05,
+      "loss": 0.5042,
+      "step": 13918
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078270304474614e-05,
+      "loss": 0.454,
+      "step": 13919
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1077156115296086e-05,
+      "loss": 0.409,
+      "step": 13920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076041912588416e-05,
+      "loss": 0.4825,
+      "step": 13921
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10749276963656e-05,
+      "loss": 0.5031,
+      "step": 13922
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1073813466641633e-05,
+      "loss": 0.421,
+      "step": 13923
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072699223430507e-05,
+      "loss": 0.4428,
+      "step": 13924
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107158496674622e-05,
+      "loss": 0.4999,
+      "step": 13925
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070470696602762e-05,
+      "loss": 0.493,
+      "step": 13926
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069356413014137e-05,
+      "loss": 0.3972,
+      "step": 13927
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1068242115994332e-05,
+      "loss": 0.4576,
+      "step": 13928
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067127805557349e-05,
+      "loss": 0.4739,
+      "step": 13929
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106601348171718e-05,
+      "loss": 0.4783,
+      "step": 13930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064899144487824e-05,
+      "loss": 0.4431,
+      "step": 13931
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106378479388327e-05,
+      "loss": 0.4474,
+      "step": 13932
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062670429917523e-05,
+      "loss": 0.4733,
+      "step": 13933
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061556052604579e-05,
+      "loss": 0.4633,
+      "step": 13934
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060441661958426e-05,
+      "loss": 0.4586,
+      "step": 13935
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059327257993066e-05,
+      "loss": 0.4965,
+      "step": 13936
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058212840722499e-05,
+      "loss": 0.4487,
+      "step": 13937
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1057098410160718e-05,
+      "loss": 0.4431,
+      "step": 13938
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055983966321726e-05,
+      "loss": 0.4676,
+      "step": 13939
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054869509219513e-05,
+      "loss": 0.4637,
+      "step": 13940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1053755038868076e-05,
+      "loss": 0.4326,
+      "step": 13941
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052640555281421e-05,
+      "loss": 0.4699,
+      "step": 13942
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1051526058473543e-05,
+      "loss": 0.438,
+      "step": 13943
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050411548458438e-05,
+      "loss": 0.3937,
+      "step": 13944
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049297025250103e-05,
+      "loss": 0.4782,
+      "step": 13945
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048182488862537e-05,
+      "loss": 0.4523,
+      "step": 13946
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047067939309746e-05,
+      "loss": 0.422,
+      "step": 13947
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045953376605721e-05,
+      "loss": 0.4203,
+      "step": 13948
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104483880076446e-05,
+      "loss": 0.4823,
+      "step": 13949
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1043724211799968e-05,
+      "loss": 0.4447,
+      "step": 13950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104260960972624e-05,
+      "loss": 0.4281,
+      "step": 13951
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1041494994557275e-05,
+      "loss": 0.4849,
+      "step": 13952
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104038036630708e-05,
+      "loss": 0.5007,
+      "step": 13953
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1039265724989645e-05,
+      "loss": 0.4338,
+      "step": 13954
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038151070618976e-05,
+      "loss": 0.4538,
+      "step": 13955
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1037036403209072e-05,
+      "loss": 0.4789,
+      "step": 13956
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1035921722773932e-05,
+      "loss": 0.5487,
+      "step": 13957
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034807029327556e-05,
+      "loss": 0.4356,
+      "step": 13958
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033692322883947e-05,
+      "loss": 0.4321,
+      "step": 13959
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1032577603457104e-05,
+      "loss": 0.45,
+      "step": 13960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103146287106103e-05,
+      "loss": 0.4159,
+      "step": 13961
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030348125709723e-05,
+      "loss": 0.4686,
+      "step": 13962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029233367417185e-05,
+      "loss": 0.4955,
+      "step": 13963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102811859619742e-05,
+      "loss": 0.4485,
+      "step": 13964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1027003812064425e-05,
+      "loss": 0.4487,
+      "step": 13965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102588901503221e-05,
+      "loss": 0.4974,
+      "step": 13966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1024774205114768e-05,
+      "loss": 0.5004,
+      "step": 13967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023659382326104e-05,
+      "loss": 0.4368,
+      "step": 13968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022544546680221e-05,
+      "loss": 0.4814,
+      "step": 13969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1021429698191122e-05,
+      "loss": 0.4646,
+      "step": 13970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1020314836872813e-05,
+      "loss": 0.5222,
+      "step": 13971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019199962739284e-05,
+      "loss": 0.476,
+      "step": 13972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018085075804552e-05,
+      "loss": 0.4572,
+      "step": 13973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016970176082613e-05,
+      "loss": 0.429,
+      "step": 13974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015855263587469e-05,
+      "loss": 0.4288,
+      "step": 13975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014740338333132e-05,
+      "loss": 0.4193,
+      "step": 13976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1013625400333596e-05,
+      "loss": 0.4247,
+      "step": 13977
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012510449602867e-05,
+      "loss": 0.4975,
+      "step": 13978
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1011395486154952e-05,
+      "loss": 0.4758,
+      "step": 13979
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101028051000385e-05,
+      "loss": 0.4127,
+      "step": 13980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009165521163574e-05,
+      "loss": 0.4725,
+      "step": 13981
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008050519648119e-05,
+      "loss": 0.4333,
+      "step": 13982
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1006935505471492e-05,
+      "loss": 0.4832,
+      "step": 13983
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10058204786477e-05,
+      "loss": 0.4624,
+      "step": 13984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004705439190746e-05,
+      "loss": 0.4602,
+      "step": 13985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1003590387114635e-05,
+      "loss": 0.432,
+      "step": 13986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002475322433374e-05,
+      "loss": 0.4258,
+      "step": 13987
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001360245160968e-05,
+      "loss": 0.4924,
+      "step": 13988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1000245155311417e-05,
+      "loss": 0.4932,
+      "step": 13989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0999130052898734e-05,
+      "loss": 0.4604,
+      "step": 13990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099801493793692e-05,
+      "loss": 0.4597,
+      "step": 13991
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996899810439984e-05,
+      "loss": 0.4221,
+      "step": 13992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099578467042193e-05,
+      "loss": 0.4755,
+      "step": 13993
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0994669517896766e-05,
+      "loss": 0.4576,
+      "step": 13994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0993554352878493e-05,
+      "loss": 0.4683,
+      "step": 13995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992439175381124e-05,
+      "loss": 0.4332,
+      "step": 13996
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0991323985418665e-05,
+      "loss": 0.4721,
+      "step": 13997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099020878300512e-05,
+      "loss": 0.4574,
+      "step": 13998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989093568154499e-05,
+      "loss": 0.4596,
+      "step": 13999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0987978340880802e-05,
+      "loss": 0.4604,
+      "step": 14000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986863101198047e-05,
+      "loss": 0.4968,
+      "step": 14001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985747849120234e-05,
+      "loss": 0.4859,
+      "step": 14002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0984632584661374e-05,
+      "loss": 0.4414,
+      "step": 14003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983517307835474e-05,
+      "loss": 0.4594,
+      "step": 14004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098240201865654e-05,
+      "loss": 0.4319,
+      "step": 14005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981286717138579e-05,
+      "loss": 0.4654,
+      "step": 14006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.5011,
+      "step": 14007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979056077141628e-05,
+      "loss": 0.4137,
+      "step": 14008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977940738690645e-05,
+      "loss": 0.4905,
+      "step": 14009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0976825387956673e-05,
+      "loss": 0.4678,
+      "step": 14010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.097571002495372e-05,
+      "loss": 0.4254,
+      "step": 14011
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0974594649695796e-05,
+      "loss": 0.4329,
+      "step": 14012
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973479262196906e-05,
+      "loss": 0.4653,
+      "step": 14013
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972363862471065e-05,
+      "loss": 0.4508,
+      "step": 14014
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971248450532275e-05,
+      "loss": 0.51,
+      "step": 14015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0970133026394553e-05,
+      "loss": 0.4418,
+      "step": 14016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0969017590071904e-05,
+      "loss": 0.4058,
+      "step": 14017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967902141578344e-05,
+      "loss": 0.4875,
+      "step": 14018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966786680927875e-05,
+      "loss": 0.4556,
+      "step": 14019
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096567120813451e-05,
+      "loss": 0.4985,
+      "step": 14020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964555723212265e-05,
+      "loss": 0.4464,
+      "step": 14021
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0963440226175144e-05,
+      "loss": 0.4956,
+      "step": 14022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962324717037158e-05,
+      "loss": 0.4629,
+      "step": 14023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0961209195812324e-05,
+      "loss": 0.4531,
+      "step": 14024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960093662514644e-05,
+      "loss": 0.4301,
+      "step": 14025
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958978117158134e-05,
+      "loss": 0.414,
+      "step": 14026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957862559756809e-05,
+      "loss": 0.4428,
+      "step": 14027
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956746990324673e-05,
+      "loss": 0.4607,
+      "step": 14028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955631408875745e-05,
+      "loss": 0.4552,
+      "step": 14029
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095451581542403e-05,
+      "loss": 0.4778,
+      "step": 14030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0953400209983544e-05,
+      "loss": 0.4843,
+      "step": 14031
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0952284592568297e-05,
+      "loss": 0.4337,
+      "step": 14032
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951168963192303e-05,
+      "loss": 0.4294,
+      "step": 14033
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950053321869576e-05,
+      "loss": 0.4653,
+      "step": 14034
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948937668614121e-05,
+      "loss": 0.4259,
+      "step": 14035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947822003439959e-05,
+      "loss": 0.4425,
+      "step": 14036
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0946706326361096e-05,
+      "loss": 0.3968,
+      "step": 14037
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0945590637391553e-05,
+      "loss": 0.3972,
+      "step": 14038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944474936545338e-05,
+      "loss": 0.4358,
+      "step": 14039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943359223836461e-05,
+      "loss": 0.495,
+      "step": 14040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942243499278944e-05,
+      "loss": 0.435,
+      "step": 14041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0941127762886793e-05,
+      "loss": 0.4904,
+      "step": 14042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940012014674024e-05,
+      "loss": 0.4422,
+      "step": 14043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0938896254654655e-05,
+      "loss": 0.4598,
+      "step": 14044
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937780482842694e-05,
+      "loss": 0.4095,
+      "step": 14045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0936664699252157e-05,
+      "loss": 0.4781,
+      "step": 14046
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093554890389706e-05,
+      "loss": 0.4993,
+      "step": 14047
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934433096791417e-05,
+      "loss": 0.4817,
+      "step": 14048
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0933317277949243e-05,
+      "loss": 0.4357,
+      "step": 14049
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932201447384549e-05,
+      "loss": 0.4671,
+      "step": 14050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0931085605111354e-05,
+      "loss": 0.418,
+      "step": 14051
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092996975114367e-05,
+      "loss": 0.452,
+      "step": 14052
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0928853885495517e-05,
+      "loss": 0.5081,
+      "step": 14053
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927738008180904e-05,
+      "loss": 0.4596,
+      "step": 14054
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926622119213853e-05,
+      "loss": 0.4492,
+      "step": 14055
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0925506218608372e-05,
+      "loss": 0.4507,
+      "step": 14056
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924390306378482e-05,
+      "loss": 0.4992,
+      "step": 14057
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09232743825382e-05,
+      "loss": 0.4407,
+      "step": 14058
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922158447101539e-05,
+      "loss": 0.4496,
+      "step": 14059
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0921042500082517e-05,
+      "loss": 0.457,
+      "step": 14060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0919926541495149e-05,
+      "loss": 0.4436,
+      "step": 14061
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918810571353452e-05,
+      "loss": 0.4432,
+      "step": 14062
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917694589671443e-05,
+      "loss": 0.393,
+      "step": 14063
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916578596463139e-05,
+      "loss": 0.4041,
+      "step": 14064
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915462591742557e-05,
+      "loss": 0.4905,
+      "step": 14065
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0914346575523713e-05,
+      "loss": 0.4492,
+      "step": 14066
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913230547820623e-05,
+      "loss": 0.4622,
+      "step": 14067
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912114508647308e-05,
+      "loss": 0.461,
+      "step": 14068
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0910998458017783e-05,
+      "loss": 0.4489,
+      "step": 14069
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0909882395946067e-05,
+      "loss": 0.4686,
+      "step": 14070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908766322446178e-05,
+      "loss": 0.4968,
+      "step": 14071
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090765023753213e-05,
+      "loss": 0.4588,
+      "step": 14072
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0906534141217949e-05,
+      "loss": 0.4358,
+      "step": 14073
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905418033517642e-05,
+      "loss": 0.4228,
+      "step": 14074
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090430191444524e-05,
+      "loss": 0.4958,
+      "step": 14075
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903185784014753e-05,
+      "loss": 0.4849,
+      "step": 14076
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09020696422402e-05,
+      "loss": 0.4942,
+      "step": 14077
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0900953489135604e-05,
+      "loss": 0.4905,
+      "step": 14078
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899837324714981e-05,
+      "loss": 0.4496,
+      "step": 14079
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089872114899235e-05,
+      "loss": 0.4368,
+      "step": 14080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897604961981733e-05,
+      "loss": 0.4407,
+      "step": 14081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896488763697149e-05,
+      "loss": 0.4925,
+      "step": 14082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895372554152611e-05,
+      "loss": 0.4809,
+      "step": 14083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894256333362145e-05,
+      "loss": 0.4835,
+      "step": 14084
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893140101339772e-05,
+      "loss": 0.478,
+      "step": 14085
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892023858099507e-05,
+      "loss": 0.4597,
+      "step": 14086
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0890907603655374e-05,
+      "loss": 0.4677,
+      "step": 14087
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088979133802139e-05,
+      "loss": 0.4953,
+      "step": 14088
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0888675061211576e-05,
+      "loss": 0.4901,
+      "step": 14089
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887558773239954e-05,
+      "loss": 0.4514,
+      "step": 14090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886442474120547e-05,
+      "loss": 0.5083,
+      "step": 14091
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885326163867372e-05,
+      "loss": 0.48,
+      "step": 14092
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088420984249445e-05,
+      "loss": 0.4574,
+      "step": 14093
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883093510015798e-05,
+      "loss": 0.4438,
+      "step": 14094
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881977166445448e-05,
+      "loss": 0.4192,
+      "step": 14095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0880860811797413e-05,
+      "loss": 0.4963,
+      "step": 14096
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879744446085721e-05,
+      "loss": 0.4262,
+      "step": 14097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878628069324386e-05,
+      "loss": 0.444,
+      "step": 14098
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877511681527433e-05,
+      "loss": 0.4562,
+      "step": 14099
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876395282708882e-05,
+      "loss": 0.4454,
+      "step": 14100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0875278872882762e-05,
+      "loss": 0.4459,
+      "step": 14101
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874162452063088e-05,
+      "loss": 0.4495,
+      "step": 14102
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873046020263886e-05,
+      "loss": 0.4384,
+      "step": 14103
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0871929577499177e-05,
+      "loss": 0.467,
+      "step": 14104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870813123782982e-05,
+      "loss": 0.4543,
+      "step": 14105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0869696659129323e-05,
+      "loss": 0.4595,
+      "step": 14106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868580183552231e-05,
+      "loss": 0.441,
+      "step": 14107
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.086746369706572e-05,
+      "loss": 0.4401,
+      "step": 14108
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866347199683818e-05,
+      "loss": 0.4349,
+      "step": 14109
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0865230691420544e-05,
+      "loss": 0.5169,
+      "step": 14110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864114172289923e-05,
+      "loss": 0.4538,
+      "step": 14111
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0862997642305983e-05,
+      "loss": 0.4299,
+      "step": 14112
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0861881101482748e-05,
+      "loss": 0.456,
+      "step": 14113
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.086076454983423e-05,
+      "loss": 0.4061,
+      "step": 14114
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0859647987374467e-05,
+      "loss": 0.4395,
+      "step": 14115
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0858531414117475e-05,
+      "loss": 0.4699,
+      "step": 14116
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857414830077278e-05,
+      "loss": 0.4012,
+      "step": 14117
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0856298235267908e-05,
+      "loss": 0.4023,
+      "step": 14118
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085518162970338e-05,
+      "loss": 0.3962,
+      "step": 14119
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854065013397723e-05,
+      "loss": 0.4914,
+      "step": 14120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852948386364963e-05,
+      "loss": 0.4873,
+      "step": 14121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851831748619121e-05,
+      "loss": 0.4534,
+      "step": 14122
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085071510017423e-05,
+      "loss": 0.4407,
+      "step": 14123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849598441044304e-05,
+      "loss": 0.4154,
+      "step": 14124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0848481771243378e-05,
+      "loss": 0.4546,
+      "step": 14125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0847365090785469e-05,
+      "loss": 0.4414,
+      "step": 14126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846248399684611e-05,
+      "loss": 0.4788,
+      "step": 14127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845131697954825e-05,
+      "loss": 0.4179,
+      "step": 14128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844014985610136e-05,
+      "loss": 0.4689,
+      "step": 14129
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842898262664573e-05,
+      "loss": 0.4879,
+      "step": 14130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841781529132158e-05,
+      "loss": 0.4821,
+      "step": 14131
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0840664785026923e-05,
+      "loss": 0.45,
+      "step": 14132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083954803036289e-05,
+      "loss": 0.476,
+      "step": 14133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838431265154089e-05,
+      "loss": 0.388,
+      "step": 14134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837314489414541e-05,
+      "loss": 0.4978,
+      "step": 14135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836197703158277e-05,
+      "loss": 0.4799,
+      "step": 14136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0835080906399322e-05,
+      "loss": 0.425,
+      "step": 14137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0833964099151706e-05,
+      "loss": 0.4352,
+      "step": 14138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832847281429456e-05,
+      "loss": 0.436,
+      "step": 14139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831730453246592e-05,
+      "loss": 0.3645,
+      "step": 14140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0830613614617152e-05,
+      "loss": 0.4351,
+      "step": 14141
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829496765555157e-05,
+      "loss": 0.4666,
+      "step": 14142
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828379906074634e-05,
+      "loss": 0.449,
+      "step": 14143
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827263036189617e-05,
+      "loss": 0.4331,
+      "step": 14144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0826146155914126e-05,
+      "loss": 0.5026,
+      "step": 14145
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825029265262193e-05,
+      "loss": 0.4554,
+      "step": 14146
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823912364247848e-05,
+      "loss": 0.439,
+      "step": 14147
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822795452885116e-05,
+      "loss": 0.4659,
+      "step": 14148
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082167853118803e-05,
+      "loss": 0.4581,
+      "step": 14149
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0820561599170617e-05,
+      "loss": 0.4822,
+      "step": 14150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0819444656846899e-05,
+      "loss": 0.4522,
+      "step": 14151
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081832770423091e-05,
+      "loss": 0.4846,
+      "step": 14152
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817210741336684e-05,
+      "loss": 0.4529,
+      "step": 14153
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816093768178244e-05,
+      "loss": 0.4524,
+      "step": 14154
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081497678476962e-05,
+      "loss": 0.3737,
+      "step": 14155
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081385979112484e-05,
+      "loss": 0.4398,
+      "step": 14156
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0812742787257938e-05,
+      "loss": 0.4897,
+      "step": 14157
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811625773182938e-05,
+      "loss": 0.4546,
+      "step": 14158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810508748913877e-05,
+      "loss": 0.4196,
+      "step": 14159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809391714464778e-05,
+      "loss": 0.4325,
+      "step": 14160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0808274669849674e-05,
+      "loss": 0.4553,
+      "step": 14161
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807157615082598e-05,
+      "loss": 0.4482,
+      "step": 14162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0806040550177571e-05,
+      "loss": 0.4744,
+      "step": 14163
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804923475148634e-05,
+      "loss": 0.4658,
+      "step": 14164
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803806390009812e-05,
+      "loss": 0.4754,
+      "step": 14165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0802689294775137e-05,
+      "loss": 0.435,
+      "step": 14166
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080157218945864e-05,
+      "loss": 0.4399,
+      "step": 14167
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080045507407435e-05,
+      "loss": 0.4327,
+      "step": 14168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799337948636299e-05,
+      "loss": 0.4501,
+      "step": 14169
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079822081315852e-05,
+      "loss": 0.4499,
+      "step": 14170
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797103667655042e-05,
+      "loss": 0.4637,
+      "step": 14171
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.07959865121399e-05,
+      "loss": 0.4415,
+      "step": 14172
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079486934662712e-05,
+      "loss": 0.4287,
+      "step": 14173
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0793752171130735e-05,
+      "loss": 0.4101,
+      "step": 14174
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079263498566478e-05,
+      "loss": 0.4889,
+      "step": 14175
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079151779024329e-05,
+      "loss": 0.4463,
+      "step": 14176
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790400584880285e-05,
+      "loss": 0.4078,
+      "step": 14177
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078928336958981e-05,
+      "loss": 0.5077,
+      "step": 14178
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0788166144385888e-05,
+      "loss": 0.4332,
+      "step": 14179
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787048909282554e-05,
+      "loss": 0.4281,
+      "step": 14180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785931664293847e-05,
+      "loss": 0.4567,
+      "step": 14181
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784814409433787e-05,
+      "loss": 0.4664,
+      "step": 14182
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078369714471642e-05,
+      "loss": 0.4533,
+      "step": 14183
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782579870155771e-05,
+      "loss": 0.4429,
+      "step": 14184
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0781462585765875e-05,
+      "loss": 0.451,
+      "step": 14185
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780345291560765e-05,
+      "loss": 0.4683,
+      "step": 14186
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779227987554475e-05,
+      "loss": 0.432,
+      "step": 14187
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778110673761035e-05,
+      "loss": 0.5249,
+      "step": 14188
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776993350194484e-05,
+      "loss": 0.4456,
+      "step": 14189
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0775876016868853e-05,
+      "loss": 0.4457,
+      "step": 14190
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774758673798178e-05,
+      "loss": 0.4631,
+      "step": 14191
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773641320996488e-05,
+      "loss": 0.4083,
+      "step": 14192
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772523958477821e-05,
+      "loss": 0.3511,
+      "step": 14193
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771406586256205e-05,
+      "loss": 0.4088,
+      "step": 14194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770289204345683e-05,
+      "loss": 0.4221,
+      "step": 14195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0769171812760286e-05,
+      "loss": 0.5045,
+      "step": 14196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076805441151405e-05,
+      "loss": 0.4515,
+      "step": 14197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766937000621002e-05,
+      "loss": 0.4125,
+      "step": 14198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0765819580095184e-05,
+      "loss": 0.509,
+      "step": 14199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764702149950632e-05,
+      "loss": 0.4591,
+      "step": 14200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0763584710201376e-05,
+      "loss": 0.4208,
+      "step": 14201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762467260861454e-05,
+      "loss": 0.4728,
+      "step": 14202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.07613498019449e-05,
+      "loss": 0.425,
+      "step": 14203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076023233346575e-05,
+      "loss": 0.4925,
+      "step": 14204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759114855438038e-05,
+      "loss": 0.4332,
+      "step": 14205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757997367875804e-05,
+      "loss": 0.4472,
+      "step": 14206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0756879870793078e-05,
+      "loss": 0.4695,
+      "step": 14207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0755762364203898e-05,
+      "loss": 0.4381,
+      "step": 14208
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0754644848122303e-05,
+      "loss": 0.4733,
+      "step": 14209
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0753527322562327e-05,
+      "loss": 0.4619,
+      "step": 14210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0752409787538e-05,
+      "loss": 0.4373,
+      "step": 14211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.075129224306337e-05,
+      "loss": 0.502,
+      "step": 14212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0750174689152467e-05,
+      "loss": 0.496,
+      "step": 14213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0749057125819325e-05,
+      "loss": 0.4831,
+      "step": 14214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747939553077984e-05,
+      "loss": 0.4746,
+      "step": 14215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746821970942481e-05,
+      "loss": 0.426,
+      "step": 14216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0745704379426849e-05,
+      "loss": 0.4416,
+      "step": 14217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744586778545135e-05,
+      "loss": 0.4511,
+      "step": 14218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743469168311362e-05,
+      "loss": 0.414,
+      "step": 14219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074235154873958e-05,
+      "loss": 0.4651,
+      "step": 14220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741233919843818e-05,
+      "loss": 0.4525,
+      "step": 14221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740116281638117e-05,
+      "loss": 0.4617,
+      "step": 14222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0738998634136518e-05,
+      "loss": 0.4718,
+      "step": 14223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0737880977353052e-05,
+      "loss": 0.4512,
+      "step": 14224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0736763311301756e-05,
+      "loss": 0.4682,
+      "step": 14225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735645635996676e-05,
+      "loss": 0.4752,
+      "step": 14226
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734527951451843e-05,
+      "loss": 0.4267,
+      "step": 14227
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.07334102576813e-05,
+      "loss": 0.4408,
+      "step": 14228
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0732292554699082e-05,
+      "loss": 0.4331,
+      "step": 14229
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073117484251923e-05,
+      "loss": 0.4419,
+      "step": 14230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730057121155778e-05,
+      "loss": 0.4885,
+      "step": 14231
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072893939062277e-05,
+      "loss": 0.488,
+      "step": 14232
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072782165093424e-05,
+      "loss": 0.4652,
+      "step": 14233
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726703902104233e-05,
+      "loss": 0.4514,
+      "step": 14234
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0725586144146782e-05,
+      "loss": 0.4044,
+      "step": 14235
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0724468377075927e-05,
+      "loss": 0.4504,
+      "step": 14236
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723350600905711e-05,
+      "loss": 0.528,
+      "step": 14237
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0722232815650171e-05,
+      "loss": 0.5103,
+      "step": 14238
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721115021323345e-05,
+      "loss": 0.452,
+      "step": 14239
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719997217939276e-05,
+      "loss": 0.3972,
+      "step": 14240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0718879405512e-05,
+      "loss": 0.4598,
+      "step": 14241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717761584055557e-05,
+      "loss": 0.4588,
+      "step": 14242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071664375358399e-05,
+      "loss": 0.4814,
+      "step": 14243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0715525914111337e-05,
+      "loss": 0.4812,
+      "step": 14244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071440806565164e-05,
+      "loss": 0.5343,
+      "step": 14245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713290208218936e-05,
+      "loss": 0.4052,
+      "step": 14246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0712172341827267e-05,
+      "loss": 0.4977,
+      "step": 14247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711054466490674e-05,
+      "loss": 0.4666,
+      "step": 14248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0709936582223195e-05,
+      "loss": 0.438,
+      "step": 14249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708818689038877e-05,
+      "loss": 0.4401,
+      "step": 14250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0707700786951754e-05,
+      "loss": 0.4577,
+      "step": 14251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070658287597587e-05,
+      "loss": 0.4755,
+      "step": 14252
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705464956125264e-05,
+      "loss": 0.4712,
+      "step": 14253
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070434702741398e-05,
+      "loss": 0.4702,
+      "step": 14254
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070322908985606e-05,
+      "loss": 0.4262,
+      "step": 14255
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702111143465538e-05,
+      "loss": 0.4131,
+      "step": 14256
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700993188256466e-05,
+      "loss": 0.4879,
+      "step": 14257
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699875224242878e-05,
+      "loss": 0.4974,
+      "step": 14258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698757251438816e-05,
+      "loss": 0.4898,
+      "step": 14259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0697639269858328e-05,
+      "loss": 0.4855,
+      "step": 14260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069652127951545e-05,
+      "loss": 0.4361,
+      "step": 14261
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695403280424224e-05,
+      "loss": 0.4563,
+      "step": 14262
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694285272598695e-05,
+      "loss": 0.4655,
+      "step": 14263
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0693167256052906e-05,
+      "loss": 0.5125,
+      "step": 14264
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692049230800894e-05,
+      "loss": 0.4609,
+      "step": 14265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690931196856706e-05,
+      "loss": 0.4898,
+      "step": 14266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0689813154234386e-05,
+      "loss": 0.4173,
+      "step": 14267
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068869510294797e-05,
+      "loss": 0.4499,
+      "step": 14268
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687577043011508e-05,
+      "loss": 0.4535,
+      "step": 14269
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0686458974439037e-05,
+      "loss": 0.4791,
+      "step": 14270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685340897244607e-05,
+      "loss": 0.4443,
+      "step": 14271
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0684222811442255e-05,
+      "loss": 0.43,
+      "step": 14272
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683104717046025e-05,
+      "loss": 0.4798,
+      "step": 14273
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681986614069961e-05,
+      "loss": 0.4949,
+      "step": 14274
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068086850252811e-05,
+      "loss": 0.512,
+      "step": 14275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067975038243451e-05,
+      "loss": 0.4242,
+      "step": 14276
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067863225380321e-05,
+      "loss": 0.4901,
+      "step": 14277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677514116648249e-05,
+      "loss": 0.5053,
+      "step": 14278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067639597098367e-05,
+      "loss": 0.4829,
+      "step": 14279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675277816823522e-05,
+      "loss": 0.4263,
+      "step": 14280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674159654181851e-05,
+      "loss": 0.4403,
+      "step": 14281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067304148307269e-05,
+      "loss": 0.4487,
+      "step": 14282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671923303510094e-05,
+      "loss": 0.4267,
+      "step": 14283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0670805115508102e-05,
+      "loss": 0.4624,
+      "step": 14284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0669686919080759e-05,
+      "loss": 0.4383,
+      "step": 14285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668568714242117e-05,
+      "loss": 0.5001,
+      "step": 14286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667450501006211e-05,
+      "loss": 0.4325,
+      "step": 14287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666332279387086e-05,
+      "loss": 0.4544,
+      "step": 14288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665214049398793e-05,
+      "loss": 0.4733,
+      "step": 14289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0664095811055373e-05,
+      "loss": 0.4491,
+      "step": 14290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662977564370878e-05,
+      "loss": 0.4965,
+      "step": 14291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661859309359344e-05,
+      "loss": 0.49,
+      "step": 14292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0660741046034818e-05,
+      "loss": 0.4569,
+      "step": 14293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659622774411347e-05,
+      "loss": 0.4592,
+      "step": 14294
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658504494502978e-05,
+      "loss": 0.499,
+      "step": 14295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0657386206323759e-05,
+      "loss": 0.4497,
+      "step": 14296
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065626790988773e-05,
+      "loss": 0.4687,
+      "step": 14297
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655149605208939e-05,
+      "loss": 0.4167,
+      "step": 14298
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.5047,
+      "step": 14299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652912971179257e-05,
+      "loss": 0.4265,
+      "step": 14300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651794641856457e-05,
+      "loss": 0.4563,
+      "step": 14301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065067630434708e-05,
+      "loss": 0.462,
+      "step": 14302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0649557958665174e-05,
+      "loss": 0.4411,
+      "step": 14303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.064843960482478e-05,
+      "loss": 0.4449,
+      "step": 14304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.064732124283995e-05,
+      "loss": 0.4599,
+      "step": 14305
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646202872724729e-05,
+      "loss": 0.4662,
+      "step": 14306
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0645084494493166e-05,
+      "loss": 0.4227,
+      "step": 14307
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643966108159303e-05,
+      "loss": 0.4806,
+      "step": 14308
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064284771373719e-05,
+      "loss": 0.4691,
+      "step": 14309
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0641729311240872e-05,
+      "loss": 0.4724,
+      "step": 14310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06406109006844e-05,
+      "loss": 0.4574,
+      "step": 14311
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0639492482081819e-05,
+      "loss": 0.4527,
+      "step": 14312
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638374055447178e-05,
+      "loss": 0.4263,
+      "step": 14313
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637255620794522e-05,
+      "loss": 0.4804,
+      "step": 14314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06361371781379e-05,
+      "loss": 0.4438,
+      "step": 14315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635018727491357e-05,
+      "loss": 0.4115,
+      "step": 14316
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633900268868944e-05,
+      "loss": 0.4898,
+      "step": 14317
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0632781802284712e-05,
+      "loss": 0.4542,
+      "step": 14318
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631663327752701e-05,
+      "loss": 0.4403,
+      "step": 14319
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0630544845286965e-05,
+      "loss": 0.5018,
+      "step": 14320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629426354901551e-05,
+      "loss": 0.4617,
+      "step": 14321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628307856610503e-05,
+      "loss": 0.4757,
+      "step": 14322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627189350427878e-05,
+      "loss": 0.4114,
+      "step": 14323
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062607083636772e-05,
+      "loss": 0.4176,
+      "step": 14324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624952314444071e-05,
+      "loss": 0.4982,
+      "step": 14325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623833784670989e-05,
+      "loss": 0.4115,
+      "step": 14326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622715247062521e-05,
+      "loss": 0.4514,
+      "step": 14327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0621596701632714e-05,
+      "loss": 0.4849,
+      "step": 14328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062047814839562e-05,
+      "loss": 0.4315,
+      "step": 14329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619359587365282e-05,
+      "loss": 0.4607,
+      "step": 14330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618241018555752e-05,
+      "loss": 0.5295,
+      "step": 14331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0617122441981084e-05,
+      "loss": 0.4564,
+      "step": 14332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0616003857655321e-05,
+      "loss": 0.4302,
+      "step": 14333
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0614885265592518e-05,
+      "loss": 0.4891,
+      "step": 14334
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061376666580672e-05,
+      "loss": 0.4635,
+      "step": 14335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612648058311976e-05,
+      "loss": 0.3824,
+      "step": 14336
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061152944312234e-05,
+      "loss": 0.4665,
+      "step": 14337
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061041082025186e-05,
+      "loss": 0.398,
+      "step": 14338
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0609292189714586e-05,
+      "loss": 0.4752,
+      "step": 14339
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0608173551524567e-05,
+      "loss": 0.4347,
+      "step": 14340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607054905695857e-05,
+      "loss": 0.4296,
+      "step": 14341
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06059362522425e-05,
+      "loss": 0.4652,
+      "step": 14342
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604817591178552e-05,
+      "loss": 0.4338,
+      "step": 14343
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060369892251806e-05,
+      "loss": 0.4184,
+      "step": 14344
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602580246275076e-05,
+      "loss": 0.4528,
+      "step": 14345
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601461562463652e-05,
+      "loss": 0.4689,
+      "step": 14346
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600342871097833e-05,
+      "loss": 0.4543,
+      "step": 14347
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599224172191675e-05,
+      "loss": 0.4158,
+      "step": 14348
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059810546575923e-05,
+      "loss": 0.5021,
+      "step": 14349
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0596986751814547e-05,
+      "loss": 0.447,
+      "step": 14350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595868030371676e-05,
+      "loss": 0.3995,
+      "step": 14351
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594749301444668e-05,
+      "loss": 0.4692,
+      "step": 14352
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0593630565047576e-05,
+      "loss": 0.4613,
+      "step": 14353
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592511821194452e-05,
+      "loss": 0.4301,
+      "step": 14354
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059139306989935e-05,
+      "loss": 0.4507,
+      "step": 14355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059027431117631e-05,
+      "loss": 0.4913,
+      "step": 14356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589155545039397e-05,
+      "loss": 0.4563,
+      "step": 14357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0588036771502655e-05,
+      "loss": 0.4948,
+      "step": 14358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586917990580136e-05,
+      "loss": 0.5149,
+      "step": 14359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05857992022859e-05,
+      "loss": 0.4462,
+      "step": 14360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584680406633991e-05,
+      "loss": 0.4302,
+      "step": 14361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0583561603638462e-05,
+      "loss": 0.4201,
+      "step": 14362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582442793313368e-05,
+      "loss": 0.4643,
+      "step": 14363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581323975672757e-05,
+      "loss": 0.459,
+      "step": 14364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058020515073069e-05,
+      "loss": 0.4506,
+      "step": 14365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579086318501209e-05,
+      "loss": 0.4835,
+      "step": 14366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0577967478998369e-05,
+      "loss": 0.4207,
+      "step": 14367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057684863223623e-05,
+      "loss": 0.4656,
+      "step": 14368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057572977822884e-05,
+      "loss": 0.5029,
+      "step": 14369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574610916990252e-05,
+      "loss": 0.4773,
+      "step": 14370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0573492048534515e-05,
+      "loss": 0.4843,
+      "step": 14371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057237317287569e-05,
+      "loss": 0.4625,
+      "step": 14372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057125429002782e-05,
+      "loss": 0.4628,
+      "step": 14373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0570135400004968e-05,
+      "loss": 0.5074,
+      "step": 14374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569016502821184e-05,
+      "loss": 0.3801,
+      "step": 14375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056789759849052e-05,
+      "loss": 0.4369,
+      "step": 14376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566778687027031e-05,
+      "loss": 0.4408,
+      "step": 14377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565659768444769e-05,
+      "loss": 0.4946,
+      "step": 14378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0564540842757789e-05,
+      "loss": 0.4092,
+      "step": 14379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0563421909980145e-05,
+      "loss": 0.4447,
+      "step": 14380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562302970125889e-05,
+      "loss": 0.4843,
+      "step": 14381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561184023209076e-05,
+      "loss": 0.4394,
+      "step": 14382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560065069243763e-05,
+      "loss": 0.4261,
+      "step": 14383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558946108243998e-05,
+      "loss": 0.5068,
+      "step": 14384
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055782714022384e-05,
+      "loss": 0.4659,
+      "step": 14385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0556708165197343e-05,
+      "loss": 0.473,
+      "step": 14386
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555589183178558e-05,
+      "loss": 0.4443,
+      "step": 14387
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554470194181546e-05,
+      "loss": 0.4684,
+      "step": 14388
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553351198220354e-05,
+      "loss": 0.4621,
+      "step": 14389
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0552232195309038e-05,
+      "loss": 0.4428,
+      "step": 14390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0551113185461657e-05,
+      "loss": 0.4643,
+      "step": 14391
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0549994168692265e-05,
+      "loss": 0.4812,
+      "step": 14392
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0548875145014912e-05,
+      "loss": 0.4853,
+      "step": 14393
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547756114443659e-05,
+      "loss": 0.4925,
+      "step": 14394
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0546637076992558e-05,
+      "loss": 0.4546,
+      "step": 14395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545518032675663e-05,
+      "loss": 0.3955,
+      "step": 14396
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0544398981507035e-05,
+      "loss": 0.4945,
+      "step": 14397
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543279923500724e-05,
+      "loss": 0.4221,
+      "step": 14398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0542160858670783e-05,
+      "loss": 0.4671,
+      "step": 14399
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541041787031275e-05,
+      "loss": 0.4592,
+      "step": 14400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0539922708596248e-05,
+      "loss": 0.4581,
+      "step": 14401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538803623379767e-05,
+      "loss": 0.4502,
+      "step": 14402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053768453139588e-05,
+      "loss": 0.4776,
+      "step": 14403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0536565432658644e-05,
+      "loss": 0.4239,
+      "step": 14404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0535446327182115e-05,
+      "loss": 0.3583,
+      "step": 14405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534327214980352e-05,
+      "loss": 0.4252,
+      "step": 14406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053320809606741e-05,
+      "loss": 0.4548,
+      "step": 14407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532088970457344e-05,
+      "loss": 0.5014,
+      "step": 14408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053096983816421e-05,
+      "loss": 0.4453,
+      "step": 14409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529850699202065e-05,
+      "loss": 0.4875,
+      "step": 14410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528731553584965e-05,
+      "loss": 0.474,
+      "step": 14411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052761240132697e-05,
+      "loss": 0.4618,
+      "step": 14412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526493242442133e-05,
+      "loss": 0.4799,
+      "step": 14413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052537407694451e-05,
+      "loss": 0.503,
+      "step": 14414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052425490484816e-05,
+      "loss": 0.4162,
+      "step": 14415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523135726167136e-05,
+      "loss": 0.439,
+      "step": 14416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05220165409155e-05,
+      "loss": 0.4728,
+      "step": 14417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520897349107309e-05,
+      "loss": 0.4862,
+      "step": 14418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0519778150756617e-05,
+      "loss": 0.4556,
+      "step": 14419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051865894587748e-05,
+      "loss": 0.4471,
+      "step": 14420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051753973448396e-05,
+      "loss": 0.4299,
+      "step": 14421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516420516590112e-05,
+      "loss": 0.4633,
+      "step": 14422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0515301292209994e-05,
+      "loss": 0.4906,
+      "step": 14423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0514182061357658e-05,
+      "loss": 0.4804,
+      "step": 14424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051306282404717e-05,
+      "loss": 0.4676,
+      "step": 14425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511943580292587e-05,
+      "loss": 0.4272,
+      "step": 14426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0510824330107959e-05,
+      "loss": 0.4255,
+      "step": 14427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0509705073507353e-05,
+      "loss": 0.4318,
+      "step": 14428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0508585810504821e-05,
+      "loss": 0.392,
+      "step": 14429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507466541114421e-05,
+      "loss": 0.4688,
+      "step": 14430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506347265350215e-05,
+      "loss": 0.5291,
+      "step": 14431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505227983226258e-05,
+      "loss": 0.4787,
+      "step": 14432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0504108694756612e-05,
+      "loss": 0.4412,
+      "step": 14433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050298939995533e-05,
+      "loss": 0.4037,
+      "step": 14434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501870098836473e-05,
+      "loss": 0.4609,
+      "step": 14435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500750791414099e-05,
+      "loss": 0.4604,
+      "step": 14436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0499631477702266e-05,
+      "loss": 0.4513,
+      "step": 14437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498512157715035e-05,
+      "loss": 0.4708,
+      "step": 14438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0497392831466464e-05,
+      "loss": 0.5237,
+      "step": 14439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496273498970613e-05,
+      "loss": 0.4577,
+      "step": 14440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495154160241534e-05,
+      "loss": 0.4463,
+      "step": 14441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494034815293294e-05,
+      "loss": 0.4391,
+      "step": 14442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049291546413995e-05,
+      "loss": 0.442,
+      "step": 14443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049179610679556e-05,
+      "loss": 0.4493,
+      "step": 14444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490676743274181e-05,
+      "loss": 0.4544,
+      "step": 14445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489557373589878e-05,
+      "loss": 0.4566,
+      "step": 14446
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488437997756703e-05,
+      "loss": 0.5065,
+      "step": 14447
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487318615788722e-05,
+      "loss": 0.4263,
+      "step": 14448
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486199227699992e-05,
+      "loss": 0.4703,
+      "step": 14449
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0485079833504572e-05,
+      "loss": 0.4491,
+      "step": 14450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0483960433216523e-05,
+      "loss": 0.4329,
+      "step": 14451
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482841026849902e-05,
+      "loss": 0.4681,
+      "step": 14452
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048172161441877e-05,
+      "loss": 0.4455,
+      "step": 14453
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480602195937189e-05,
+      "loss": 0.4488,
+      "step": 14454
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479482771419218e-05,
+      "loss": 0.4424,
+      "step": 14455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0478363340878916e-05,
+      "loss": 0.4237,
+      "step": 14456
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477243904330343e-05,
+      "loss": 0.4568,
+      "step": 14457
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047612446178756e-05,
+      "loss": 0.434,
+      "step": 14458
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475005013264628e-05,
+      "loss": 0.4658,
+      "step": 14459
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0473885558775608e-05,
+      "loss": 0.4867,
+      "step": 14460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472766098334555e-05,
+      "loss": 0.4634,
+      "step": 14461
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471646631955536e-05,
+      "loss": 0.501,
+      "step": 14462
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0470527159652608e-05,
+      "loss": 0.4298,
+      "step": 14463
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469407681439831e-05,
+      "loss": 0.4641,
+      "step": 14464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468288197331272e-05,
+      "loss": 0.4604,
+      "step": 14465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0467168707340985e-05,
+      "loss": 0.4634,
+      "step": 14466
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046604921148303e-05,
+      "loss": 0.456,
+      "step": 14467
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0464929709771471e-05,
+      "loss": 0.4378,
+      "step": 14468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046381020222037e-05,
+      "loss": 0.4453,
+      "step": 14469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0462690688843788e-05,
+      "loss": 0.4501,
+      "step": 14470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461571169655787e-05,
+      "loss": 0.526,
+      "step": 14471
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0460451644670422e-05,
+      "loss": 0.4455,
+      "step": 14472
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045933211390176e-05,
+      "loss": 0.4748,
+      "step": 14473
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045821257736386e-05,
+      "loss": 0.471,
+      "step": 14474
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0457093035070787e-05,
+      "loss": 0.4808,
+      "step": 14475
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455973487036596e-05,
+      "loss": 0.5096,
+      "step": 14476
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454853933275357e-05,
+      "loss": 0.3565,
+      "step": 14477
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0453734373801123e-05,
+      "loss": 0.4857,
+      "step": 14478
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045261480862796e-05,
+      "loss": 0.472,
+      "step": 14479
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451495237769934e-05,
+      "loss": 0.4315,
+      "step": 14480
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450375661241099e-05,
+      "loss": 0.4311,
+      "step": 14481
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449256079055521e-05,
+      "loss": 0.4169,
+      "step": 14482
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0448136491227263e-05,
+      "loss": 0.4295,
+      "step": 14483
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447016897770384e-05,
+      "loss": 0.4173,
+      "step": 14484
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0445897298698947e-05,
+      "loss": 0.4509,
+      "step": 14485
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444777694027017e-05,
+      "loss": 0.4404,
+      "step": 14486
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0443658083768653e-05,
+      "loss": 0.4761,
+      "step": 14487
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044253846793792e-05,
+      "loss": 0.4625,
+      "step": 14488
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0441418846548879e-05,
+      "loss": 0.4612,
+      "step": 14489
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440299219615588e-05,
+      "loss": 0.406,
+      "step": 14490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043917958715212e-05,
+      "loss": 0.4547,
+      "step": 14491
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0438059949172527e-05,
+      "loss": 0.4222,
+      "step": 14492
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436940305690882e-05,
+      "loss": 0.4715,
+      "step": 14493
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0435820656721238e-05,
+      "loss": 0.4271,
+      "step": 14494
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434701002277663e-05,
+      "loss": 0.4217,
+      "step": 14495
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043358134237422e-05,
+      "loss": 0.4168,
+      "step": 14496
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0432461677024972e-05,
+      "loss": 0.4837,
+      "step": 14497
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0431342006243979e-05,
+      "loss": 0.4842,
+      "step": 14498
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0430222330045306e-05,
+      "loss": 0.4427,
+      "step": 14499
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0429102648443017e-05,
+      "loss": 0.4303,
+      "step": 14500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0427982961451172e-05,
+      "loss": 0.4428,
+      "step": 14501
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0426863269083845e-05,
+      "loss": 0.4248,
+      "step": 14502
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425743571355087e-05,
+      "loss": 0.4687,
+      "step": 14503
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424623868278963e-05,
+      "loss": 0.4099,
+      "step": 14504
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423504159869543e-05,
+      "loss": 0.4988,
+      "step": 14505
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422384446140885e-05,
+      "loss": 0.4838,
+      "step": 14506
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0421264727107056e-05,
+      "loss": 0.3962,
+      "step": 14507
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0420145002782119e-05,
+      "loss": 0.4394,
+      "step": 14508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0419025273180134e-05,
+      "loss": 0.5399,
+      "step": 14509
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0417905538315172e-05,
+      "loss": 0.4744,
+      "step": 14510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041678579820129e-05,
+      "loss": 0.4352,
+      "step": 14511
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415666052852558e-05,
+      "loss": 0.4706,
+      "step": 14512
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0414546302283036e-05,
+      "loss": 0.4472,
+      "step": 14513
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413426546506788e-05,
+      "loss": 0.4797,
+      "step": 14514
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412306785537881e-05,
+      "loss": 0.4195,
+      "step": 14515
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0411187019390374e-05,
+      "loss": 0.4891,
+      "step": 14516
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0410067248078339e-05,
+      "loss": 0.5246,
+      "step": 14517
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408947471615836e-05,
+      "loss": 0.4536,
+      "step": 14518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040782769001693e-05,
+      "loss": 0.425,
+      "step": 14519
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406707903295685e-05,
+      "loss": 0.4527,
+      "step": 14520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0405588111466163e-05,
+      "loss": 0.4759,
+      "step": 14521
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404468314542435e-05,
+      "loss": 0.4519,
+      "step": 14522
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040334851253856e-05,
+      "loss": 0.4767,
+      "step": 14523
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402228705468609e-05,
+      "loss": 0.4417,
+      "step": 14524
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0401108893346638e-05,
+      "loss": 0.4211,
+      "step": 14525
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399989076186721e-05,
+      "loss": 0.4878,
+      "step": 14526
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398869254002914e-05,
+      "loss": 0.4615,
+      "step": 14527
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397749426809292e-05,
+      "loss": 0.4125,
+      "step": 14528
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396629594619913e-05,
+      "loss": 0.4817,
+      "step": 14529
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395509757448842e-05,
+      "loss": 0.4866,
+      "step": 14530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039438991531015e-05,
+      "loss": 0.4861,
+      "step": 14531
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0393270068217896e-05,
+      "loss": 0.4593,
+      "step": 14532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0392150216186147e-05,
+      "loss": 0.4448,
+      "step": 14533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0391030359228975e-05,
+      "loss": 0.4286,
+      "step": 14534
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0389910497360434e-05,
+      "loss": 0.4698,
+      "step": 14535
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388790630594596e-05,
+      "loss": 0.4527,
+      "step": 14536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0387670758945529e-05,
+      "loss": 0.5205,
+      "step": 14537
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0386550882427291e-05,
+      "loss": 0.4368,
+      "step": 14538
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385431001053958e-05,
+      "loss": 0.4315,
+      "step": 14539
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384311114839587e-05,
+      "loss": 0.4761,
+      "step": 14540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383191223798246e-05,
+      "loss": 0.4152,
+      "step": 14541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382071327944e-05,
+      "loss": 0.447,
+      "step": 14542
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380951427290922e-05,
+      "loss": 0.4748,
+      "step": 14543
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037983152185307e-05,
+      "loss": 0.4328,
+      "step": 14544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0378711611644512e-05,
+      "loss": 0.5372,
+      "step": 14545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377591696679314e-05,
+      "loss": 0.4701,
+      "step": 14546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0376471776971546e-05,
+      "loss": 0.4532,
+      "step": 14547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037535185253527e-05,
+      "loss": 0.4739,
+      "step": 14548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0374231923384554e-05,
+      "loss": 0.4794,
+      "step": 14549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373111989533465e-05,
+      "loss": 0.4173,
+      "step": 14550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371992050996067e-05,
+      "loss": 0.4537,
+      "step": 14551
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370872107786428e-05,
+      "loss": 0.5002,
+      "step": 14552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0369752159918617e-05,
+      "loss": 0.4705,
+      "step": 14553
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368632207406695e-05,
+      "loss": 0.4622,
+      "step": 14554
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0367512250264732e-05,
+      "loss": 0.4825,
+      "step": 14555
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366392288506797e-05,
+      "loss": 0.4535,
+      "step": 14556
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365272322146952e-05,
+      "loss": 0.4924,
+      "step": 14557
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0364152351199266e-05,
+      "loss": 0.4618,
+      "step": 14558
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363032375677808e-05,
+      "loss": 0.5038,
+      "step": 14559
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361912395596646e-05,
+      "loss": 0.4605,
+      "step": 14560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0360792410969838e-05,
+      "loss": 0.4006,
+      "step": 14561
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035967242181146e-05,
+      "loss": 0.4787,
+      "step": 14562
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0358552428135576e-05,
+      "loss": 0.497,
+      "step": 14563
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0357432429956251e-05,
+      "loss": 0.41,
+      "step": 14564
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0356312427287558e-05,
+      "loss": 0.4759,
+      "step": 14565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355192420143558e-05,
+      "loss": 0.4707,
+      "step": 14566
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354072408538325e-05,
+      "loss": 0.4926,
+      "step": 14567
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035295239248592e-05,
+      "loss": 0.4032,
+      "step": 14568
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0351832372000411e-05,
+      "loss": 0.4381,
+      "step": 14569
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350712347095876e-05,
+      "loss": 0.4134,
+      "step": 14570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0349592317786368e-05,
+      "loss": 0.4697,
+      "step": 14571
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034847228408596e-05,
+      "loss": 0.4198,
+      "step": 14572
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347352246008721e-05,
+      "loss": 0.4286,
+      "step": 14573
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346232203568723e-05,
+      "loss": 0.4398,
+      "step": 14574
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345112156780023e-05,
+      "loss": 0.4455,
+      "step": 14575
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0343992105656699e-05,
+      "loss": 0.429,
+      "step": 14576
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0342872050212815e-05,
+      "loss": 0.4519,
+      "step": 14577
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341751990462436e-05,
+      "loss": 0.4476,
+      "step": 14578
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0340631926419635e-05,
+      "loss": 0.4765,
+      "step": 14579
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339511858098477e-05,
+      "loss": 0.4485,
+      "step": 14580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338391785513032e-05,
+      "loss": 0.4795,
+      "step": 14581
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0337271708677368e-05,
+      "loss": 0.4134,
+      "step": 14582
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0336151627605549e-05,
+      "loss": 0.5099,
+      "step": 14583
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335031542311648e-05,
+      "loss": 0.4498,
+      "step": 14584
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0333911452809732e-05,
+      "loss": 0.5098,
+      "step": 14585
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332791359113871e-05,
+      "loss": 0.4578,
+      "step": 14586
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331671261238133e-05,
+      "loss": 0.4302,
+      "step": 14587
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0330551159196585e-05,
+      "loss": 0.4519,
+      "step": 14588
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329431053003291e-05,
+      "loss": 0.4411,
+      "step": 14589
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0328310942672328e-05,
+      "loss": 0.4623,
+      "step": 14590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327190828217763e-05,
+      "loss": 0.4399,
+      "step": 14591
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326070709653662e-05,
+      "loss": 0.4682,
+      "step": 14592
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0324950586994093e-05,
+      "loss": 0.4692,
+      "step": 14593
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0323830460253127e-05,
+      "loss": 0.4225,
+      "step": 14594
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0322710329444831e-05,
+      "loss": 0.4516,
+      "step": 14595
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0321590194583277e-05,
+      "loss": 0.4719,
+      "step": 14596
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320470055682535e-05,
+      "loss": 0.4864,
+      "step": 14597
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0319349912756667e-05,
+      "loss": 0.3894,
+      "step": 14598
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318229765819745e-05,
+      "loss": 0.4384,
+      "step": 14599
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317109614885843e-05,
+      "loss": 0.4896,
+      "step": 14600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0315989459969022e-05,
+      "loss": 0.5191,
+      "step": 14601
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031486930108336e-05,
+      "loss": 0.4405,
+      "step": 14602
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031374913824292e-05,
+      "loss": 0.495,
+      "step": 14603
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312628971461776e-05,
+      "loss": 0.4594,
+      "step": 14604
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031150880075399e-05,
+      "loss": 0.4584,
+      "step": 14605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310388626133639e-05,
+      "loss": 0.4297,
+      "step": 14606
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030926844761479e-05,
+      "loss": 0.4795,
+      "step": 14607
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308148265211513e-05,
+      "loss": 0.5241,
+      "step": 14608
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0307028078937871e-05,
+      "loss": 0.446,
+      "step": 14609
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305907888807945e-05,
+      "loss": 0.4956,
+      "step": 14610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0304787694835797e-05,
+      "loss": 0.4762,
+      "step": 14611
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303667497035497e-05,
+      "loss": 0.3786,
+      "step": 14612
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302547295421116e-05,
+      "loss": 0.4542,
+      "step": 14613
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301427090006725e-05,
+      "loss": 0.4453,
+      "step": 14614
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0300306880806392e-05,
+      "loss": 0.4604,
+      "step": 14615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299186667834189e-05,
+      "loss": 0.4754,
+      "step": 14616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0298066451104185e-05,
+      "loss": 0.3874,
+      "step": 14617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029694623063045e-05,
+      "loss": 0.4895,
+      "step": 14618
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295826006427052e-05,
+      "loss": 0.4223,
+      "step": 14619
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0294705778508063e-05,
+      "loss": 0.4754,
+      "step": 14620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293585546887552e-05,
+      "loss": 0.4579,
+      "step": 14621
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292465311579593e-05,
+      "loss": 0.4774,
+      "step": 14622
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291345072598253e-05,
+      "loss": 0.4251,
+      "step": 14623
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290224829957602e-05,
+      "loss": 0.4925,
+      "step": 14624
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028910458367171e-05,
+      "loss": 0.4412,
+      "step": 14625
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287984333754646e-05,
+      "loss": 0.4764,
+      "step": 14626
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286864080220486e-05,
+      "loss": 0.5228,
+      "step": 14627
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285743823083295e-05,
+      "loss": 0.4971,
+      "step": 14628
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028462356235715e-05,
+      "loss": 0.4695,
+      "step": 14629
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0283503298056114e-05,
+      "loss": 0.4376,
+      "step": 14630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028238303019426e-05,
+      "loss": 0.4969,
+      "step": 14631
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028126275878566e-05,
+      "loss": 0.5156,
+      "step": 14632
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280142483844384e-05,
+      "loss": 0.52,
+      "step": 14633
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279022205384503e-05,
+      "loss": 0.471,
+      "step": 14634
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277901923420086e-05,
+      "loss": 0.4872,
+      "step": 14635
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276781637965208e-05,
+      "loss": 0.4465,
+      "step": 14636
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275661349033935e-05,
+      "loss": 0.4692,
+      "step": 14637
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027454105664034e-05,
+      "loss": 0.394,
+      "step": 14638
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0273420760798495e-05,
+      "loss": 0.4869,
+      "step": 14639
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027230046152247e-05,
+      "loss": 0.4372,
+      "step": 14640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271180158826334e-05,
+      "loss": 0.4538,
+      "step": 14641
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270059852724163e-05,
+      "loss": 0.4035,
+      "step": 14642
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0268939543230021e-05,
+      "loss": 0.5133,
+      "step": 14643
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026781923035799e-05,
+      "loss": 0.4743,
+      "step": 14644
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266698914122132e-05,
+      "loss": 0.4484,
+      "step": 14645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265578594536515e-05,
+      "loss": 0.4375,
+      "step": 14646
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0264458271615221e-05,
+      "loss": 0.4697,
+      "step": 14647
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0263337945372315e-05,
+      "loss": 0.5085,
+      "step": 14648
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026221761582187e-05,
+      "loss": 0.4522,
+      "step": 14649
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261097282977959e-05,
+      "loss": 0.4096,
+      "step": 14650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259976946854648e-05,
+      "loss": 0.4721,
+      "step": 14651
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0258856607466013e-05,
+      "loss": 0.4562,
+      "step": 14652
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257736264826125e-05,
+      "loss": 0.4954,
+      "step": 14653
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0256615918949055e-05,
+      "loss": 0.449,
+      "step": 14654
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255495569848874e-05,
+      "loss": 0.3989,
+      "step": 14655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254375217539656e-05,
+      "loss": 0.4392,
+      "step": 14656
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.025325486203547e-05,
+      "loss": 0.4527,
+      "step": 14657
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252134503350389e-05,
+      "loss": 0.5244,
+      "step": 14658
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251014141498486e-05,
+      "loss": 0.5008,
+      "step": 14659
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024989377649383e-05,
+      "loss": 0.467,
+      "step": 14660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248773408350491e-05,
+      "loss": 0.459,
+      "step": 14661
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024765303708255e-05,
+      "loss": 0.4474,
+      "step": 14662
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246532662704067e-05,
+      "loss": 0.4643,
+      "step": 14663
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245412285229124e-05,
+      "loss": 0.4478,
+      "step": 14664
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244291904671788e-05,
+      "loss": 0.475,
+      "step": 14665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0243171521046132e-05,
+      "loss": 0.4532,
+      "step": 14666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242051134366228e-05,
+      "loss": 0.478,
+      "step": 14667
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024093074464615e-05,
+      "loss": 0.458,
+      "step": 14668
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239810351899964e-05,
+      "loss": 0.4696,
+      "step": 14669
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023868995614175e-05,
+      "loss": 0.4897,
+      "step": 14670
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237569557385577e-05,
+      "loss": 0.458,
+      "step": 14671
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0236449155645514e-05,
+      "loss": 0.4777,
+      "step": 14672
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0235328750935638e-05,
+      "loss": 0.5019,
+      "step": 14673
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234208343270019e-05,
+      "loss": 0.5581,
+      "step": 14674
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0233087932662731e-05,
+      "loss": 0.4381,
+      "step": 14675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0231967519127848e-05,
+      "loss": 0.4275,
+      "step": 14676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230847102679435e-05,
+      "loss": 0.3813,
+      "step": 14677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229726683331573e-05,
+      "loss": 0.5197,
+      "step": 14678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022860626109833e-05,
+      "loss": 0.4143,
+      "step": 14679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227485835993777e-05,
+      "loss": 0.4613,
+      "step": 14680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226365408031995e-05,
+      "loss": 0.4835,
+      "step": 14681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022524497722705e-05,
+      "loss": 0.4124,
+      "step": 14682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0224124543593011e-05,
+      "loss": 0.4625,
+      "step": 14683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223004107143957e-05,
+      "loss": 0.4128,
+      "step": 14684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0221883667893959e-05,
+      "loss": 0.5011,
+      "step": 14685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022076322585709e-05,
+      "loss": 0.4678,
+      "step": 14686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0219642781047423e-05,
+      "loss": 0.4694,
+      "step": 14687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218522333479029e-05,
+      "loss": 0.4639,
+      "step": 14688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0217401883165981e-05,
+      "loss": 0.4255,
+      "step": 14689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0216281430122357e-05,
+      "loss": 0.4313,
+      "step": 14690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215160974362224e-05,
+      "loss": 0.4447,
+      "step": 14691
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0214040515899657e-05,
+      "loss": 0.3854,
+      "step": 14692
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.021292005474873e-05,
+      "loss": 0.4445,
+      "step": 14693
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211799590923513e-05,
+      "loss": 0.4566,
+      "step": 14694
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0210679124438082e-05,
+      "loss": 0.4153,
+      "step": 14695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020955865530651e-05,
+      "loss": 0.4902,
+      "step": 14696
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0208438183542872e-05,
+      "loss": 0.4814,
+      "step": 14697
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207317709161238e-05,
+      "loss": 0.4647,
+      "step": 14698
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020619723217568e-05,
+      "loss": 0.4669,
+      "step": 14699
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205076752600272e-05,
+      "loss": 0.4652,
+      "step": 14700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0203956270449091e-05,
+      "loss": 0.4649,
+      "step": 14701
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020283578573621e-05,
+      "loss": 0.4644,
+      "step": 14702
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201715298475693e-05,
+      "loss": 0.4212,
+      "step": 14703
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0200594808681627e-05,
+      "loss": 0.4948,
+      "step": 14704
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199474316368076e-05,
+      "loss": 0.4844,
+      "step": 14705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0198353821549116e-05,
+      "loss": 0.4596,
+      "step": 14706
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197233324238825e-05,
+      "loss": 0.4104,
+      "step": 14707
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0196112824451269e-05,
+      "loss": 0.4201,
+      "step": 14708
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0194992322200523e-05,
+      "loss": 0.4905,
+      "step": 14709
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0193871817500667e-05,
+      "loss": 0.4406,
+      "step": 14710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192751310365765e-05,
+      "loss": 0.4452,
+      "step": 14711
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191630800809901e-05,
+      "loss": 0.4551,
+      "step": 14712
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190510288847142e-05,
+      "loss": 0.467,
+      "step": 14713
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189389774491561e-05,
+      "loss": 0.4816,
+      "step": 14714
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188269257757236e-05,
+      "loss": 0.4125,
+      "step": 14715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187148738658237e-05,
+      "loss": 0.5196,
+      "step": 14716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186028217208642e-05,
+      "loss": 0.4479,
+      "step": 14717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0184907693422518e-05,
+      "loss": 0.3666,
+      "step": 14718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0183787167313947e-05,
+      "loss": 0.5105,
+      "step": 14719
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182666638896996e-05,
+      "loss": 0.4964,
+      "step": 14720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181546108185743e-05,
+      "loss": 0.4561,
+      "step": 14721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0180425575194262e-05,
+      "loss": 0.3836,
+      "step": 14722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0179305039936624e-05,
+      "loss": 0.5321,
+      "step": 14723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0178184502426905e-05,
+      "loss": 0.4797,
+      "step": 14724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0177063962679179e-05,
+      "loss": 0.4048,
+      "step": 14725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175943420707517e-05,
+      "loss": 0.5128,
+      "step": 14726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0174822876525999e-05,
+      "loss": 0.4527,
+      "step": 14727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173702330148695e-05,
+      "loss": 0.4584,
+      "step": 14728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017258178158968e-05,
+      "loss": 0.4341,
+      "step": 14729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171461230863028e-05,
+      "loss": 0.5123,
+      "step": 14730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170340677982812e-05,
+      "loss": 0.4458,
+      "step": 14731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169220122963111e-05,
+      "loss": 0.4013,
+      "step": 14732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0168099565817994e-05,
+      "loss": 0.4661,
+      "step": 14733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166979006561538e-05,
+      "loss": 0.4291,
+      "step": 14734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0165858445207814e-05,
+      "loss": 0.4441,
+      "step": 14735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01647378817709e-05,
+      "loss": 0.4304,
+      "step": 14736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0163617316264869e-05,
+      "loss": 0.4485,
+      "step": 14737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162496748703795e-05,
+      "loss": 0.4878,
+      "step": 14738
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161376179101754e-05,
+      "loss": 0.4361,
+      "step": 14739
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160255607472818e-05,
+      "loss": 0.5173,
+      "step": 14740
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159135033831063e-05,
+      "loss": 0.4228,
+      "step": 14741
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158014458190562e-05,
+      "loss": 0.449,
+      "step": 14742
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0156893880565391e-05,
+      "loss": 0.4354,
+      "step": 14743
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155773300969625e-05,
+      "loss": 0.4336,
+      "step": 14744
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0154652719417337e-05,
+      "loss": 0.4957,
+      "step": 14745
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01535321359226e-05,
+      "loss": 0.4915,
+      "step": 14746
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152411550499495e-05,
+      "loss": 0.4576,
+      "step": 14747
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0151290963162087e-05,
+      "loss": 0.4727,
+      "step": 14748
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150170373924464e-05,
+      "loss": 0.4502,
+      "step": 14749
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149049782800688e-05,
+      "loss": 0.4815,
+      "step": 14750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147929189804834e-05,
+      "loss": 0.4408,
+      "step": 14751
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146808594950988e-05,
+      "loss": 0.4625,
+      "step": 14752
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0145687998253215e-05,
+      "loss": 0.4788,
+      "step": 14753
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144567399725594e-05,
+      "loss": 0.4209,
+      "step": 14754
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143446799382197e-05,
+      "loss": 0.4048,
+      "step": 14755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142326197237103e-05,
+      "loss": 0.3928,
+      "step": 14756
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0141205593304378e-05,
+      "loss": 0.4489,
+      "step": 14757
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0140084987598106e-05,
+      "loss": 0.4484,
+      "step": 14758
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013896438013236e-05,
+      "loss": 0.4528,
+      "step": 14759
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137843770921212e-05,
+      "loss": 0.4924,
+      "step": 14760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013672315997874e-05,
+      "loss": 0.4345,
+      "step": 14761
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135602547319016e-05,
+      "loss": 0.4731,
+      "step": 14762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134481932956117e-05,
+      "loss": 0.5008,
+      "step": 14763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133361316904118e-05,
+      "loss": 0.4796,
+      "step": 14764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0132240699177092e-05,
+      "loss": 0.4667,
+      "step": 14765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0131120079789116e-05,
+      "loss": 0.4437,
+      "step": 14766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129999458754264e-05,
+      "loss": 0.5087,
+      "step": 14767
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128878836086614e-05,
+      "loss": 0.4907,
+      "step": 14768
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127758211800236e-05,
+      "loss": 0.4505,
+      "step": 14769
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126637585909207e-05,
+      "loss": 0.4429,
+      "step": 14770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125516958427606e-05,
+      "loss": 0.4601,
+      "step": 14771
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0124396329369503e-05,
+      "loss": 0.461,
+      "step": 14772
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123275698748976e-05,
+      "loss": 0.4914,
+      "step": 14773
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122155066580098e-05,
+      "loss": 0.4072,
+      "step": 14774
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121034432876946e-05,
+      "loss": 0.4898,
+      "step": 14775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119913797653598e-05,
+      "loss": 0.4847,
+      "step": 14776
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118793160924122e-05,
+      "loss": 0.4178,
+      "step": 14777
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0117672522702597e-05,
+      "loss": 0.4728,
+      "step": 14778
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116551883003101e-05,
+      "loss": 0.484,
+      "step": 14779
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0115431241839702e-05,
+      "loss": 0.4553,
+      "step": 14780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0114310599226487e-05,
+      "loss": 0.4752,
+      "step": 14781
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011318995517752e-05,
+      "loss": 0.4576,
+      "step": 14782
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011206930970688e-05,
+      "loss": 0.4456,
+      "step": 14783
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0110948662828646e-05,
+      "loss": 0.5184,
+      "step": 14784
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109828014556889e-05,
+      "loss": 0.4709,
+      "step": 14785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010870736490569e-05,
+      "loss": 0.5176,
+      "step": 14786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0107586713889117e-05,
+      "loss": 0.4726,
+      "step": 14787
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0106466061521247e-05,
+      "loss": 0.4432,
+      "step": 14788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0105345407816158e-05,
+      "loss": 0.4761,
+      "step": 14789
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104224752787928e-05,
+      "loss": 0.4634,
+      "step": 14790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0103104096450626e-05,
+      "loss": 0.4611,
+      "step": 14791
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101983438818335e-05,
+      "loss": 0.4466,
+      "step": 14792
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0100862779905123e-05,
+      "loss": 0.4592,
+      "step": 14793
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0099742119725067e-05,
+      "loss": 0.433,
+      "step": 14794
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098621458292246e-05,
+      "loss": 0.4836,
+      "step": 14795
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097500795620735e-05,
+      "loss": 0.442,
+      "step": 14796
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009638013172461e-05,
+      "loss": 0.4453,
+      "step": 14797
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095259466617944e-05,
+      "loss": 0.4404,
+      "step": 14798
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094138800314812e-05,
+      "loss": 0.4592,
+      "step": 14799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093018132829292e-05,
+      "loss": 0.4726,
+      "step": 14800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009189746417546e-05,
+      "loss": 0.4701,
+      "step": 14801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090776794367389e-05,
+      "loss": 0.4249,
+      "step": 14802
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089656123419159e-05,
+      "loss": 0.4999,
+      "step": 14803
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008853545134484e-05,
+      "loss": 0.473,
+      "step": 14804
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087414778158512e-05,
+      "loss": 0.5089,
+      "step": 14805
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008629410387425e-05,
+      "loss": 0.4671,
+      "step": 14806
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008517342850613e-05,
+      "loss": 0.4079,
+      "step": 14807
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084052752068225e-05,
+      "loss": 0.4972,
+      "step": 14808
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0082932074574616e-05,
+      "loss": 0.4221,
+      "step": 14809
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0081811396039374e-05,
+      "loss": 0.4011,
+      "step": 14810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080690716476573e-05,
+      "loss": 0.3891,
+      "step": 14811
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079570035900294e-05,
+      "loss": 0.4298,
+      "step": 14812
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078449354324616e-05,
+      "loss": 0.4271,
+      "step": 14813
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0077328671763601e-05,
+      "loss": 0.4215,
+      "step": 14814
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007620798823134e-05,
+      "loss": 0.4705,
+      "step": 14815
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0075087303741899e-05,
+      "loss": 0.4427,
+      "step": 14816
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007396661830936e-05,
+      "loss": 0.4611,
+      "step": 14817
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0072845931947795e-05,
+      "loss": 0.457,
+      "step": 14818
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071725244671281e-05,
+      "loss": 0.4815,
+      "step": 14819
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070604556493893e-05,
+      "loss": 0.4827,
+      "step": 14820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069483867429708e-05,
+      "loss": 0.4671,
+      "step": 14821
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068363177492802e-05,
+      "loss": 0.423,
+      "step": 14822
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0067242486697251e-05,
+      "loss": 0.4715,
+      "step": 14823
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066121795057131e-05,
+      "loss": 0.4739,
+      "step": 14824
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065001102586515e-05,
+      "loss": 0.4495,
+      "step": 14825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063880409299485e-05,
+      "loss": 0.469,
+      "step": 14826
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006275971521011e-05,
+      "loss": 0.4246,
+      "step": 14827
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0061639020332472e-05,
+      "loss": 0.4754,
+      "step": 14828
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060518324680647e-05,
+      "loss": 0.4119,
+      "step": 14829
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059397628268703e-05,
+      "loss": 0.4335,
+      "step": 14830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058276931110723e-05,
+      "loss": 0.4643,
+      "step": 14831
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057156233220782e-05,
+      "loss": 0.4463,
+      "step": 14832
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056035534612955e-05,
+      "loss": 0.4635,
+      "step": 14833
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054914835301322e-05,
+      "loss": 0.4821,
+      "step": 14834
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053794135299953e-05,
+      "loss": 0.4743,
+      "step": 14835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0052673434622926e-05,
+      "loss": 0.539,
+      "step": 14836
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051552733284318e-05,
+      "loss": 0.4958,
+      "step": 14837
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050432031298205e-05,
+      "loss": 0.4563,
+      "step": 14838
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004931132867866e-05,
+      "loss": 0.5032,
+      "step": 14839
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0048190625439766e-05,
+      "loss": 0.4838,
+      "step": 14840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047069921595595e-05,
+      "loss": 0.4702,
+      "step": 14841
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004594921716022e-05,
+      "loss": 0.4283,
+      "step": 14842
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044828512147723e-05,
+      "loss": 0.4567,
+      "step": 14843
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004370780657218e-05,
+      "loss": 0.5127,
+      "step": 14844
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042587100447659e-05,
+      "loss": 0.4525,
+      "step": 14845
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041466393788244e-05,
+      "loss": 0.4936,
+      "step": 14846
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0040345686608008e-05,
+      "loss": 0.4026,
+      "step": 14847
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039224978921026e-05,
+      "loss": 0.5154,
+      "step": 14848
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038104270741382e-05,
+      "loss": 0.4393,
+      "step": 14849
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036983562083145e-05,
+      "loss": 0.412,
+      "step": 14850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0035862852960388e-05,
+      "loss": 0.4271,
+      "step": 14851
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034742143387193e-05,
+      "loss": 0.4848,
+      "step": 14852
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0033621433377635e-05,
+      "loss": 0.4582,
+      "step": 14853
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032500722945793e-05,
+      "loss": 0.4688,
+      "step": 14854
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0031380012105738e-05,
+      "loss": 0.4715,
+      "step": 14855
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030259300871545e-05,
+      "loss": 0.5106,
+      "step": 14856
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029138589257297e-05,
+      "loss": 0.4437,
+      "step": 14857
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0028017877277066e-05,
+      "loss": 0.3915,
+      "step": 14858
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026897164944929e-05,
+      "loss": 0.4273,
+      "step": 14859
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025776452274962e-05,
+      "loss": 0.44,
+      "step": 14860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0024655739281242e-05,
+      "loss": 0.4621,
+      "step": 14861
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023535025977842e-05,
+      "loss": 0.4252,
+      "step": 14862
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022414312378843e-05,
+      "loss": 0.4046,
+      "step": 14863
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021293598498318e-05,
+      "loss": 0.4297,
+      "step": 14864
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020172884350345e-05,
+      "loss": 0.4244,
+      "step": 14865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019052169949e-05,
+      "loss": 0.4563,
+      "step": 14866
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017931455308359e-05,
+      "loss": 0.4998,
+      "step": 14867
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016810740442495e-05,
+      "loss": 0.4123,
+      "step": 14868
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015690025365488e-05,
+      "loss": 0.4824,
+      "step": 14869
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014569310091416e-05,
+      "loss": 0.4549,
+      "step": 14870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013448594634352e-05,
+      "loss": 0.5158,
+      "step": 14871
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001232787900837e-05,
+      "loss": 0.4488,
+      "step": 14872
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011207163227553e-05,
+      "loss": 0.4786,
+      "step": 14873
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001008644730597e-05,
+      "loss": 0.4232,
+      "step": 14874
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00089657312577e-05,
+      "loss": 0.4339,
+      "step": 14875
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007845015096825e-05,
+      "loss": 0.4647,
+      "step": 14876
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000672429883741e-05,
+      "loss": 0.5024,
+      "step": 14877
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005603582493544e-05,
+      "loss": 0.4595,
+      "step": 14878
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000448286607929e-05,
+      "loss": 0.4894,
+      "step": 14879
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003362149608735e-05,
+      "loss": 0.4847,
+      "step": 14880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0002241433095953e-05,
+      "loss": 0.4363,
+      "step": 14881
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0001120716555016e-05,
+      "loss": 0.4317,
+      "step": 14882
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.4259,
+      "step": 14883
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998879283444986e-06,
+      "loss": 0.448,
+      "step": 14884
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997758566904052e-06,
+      "loss": 0.4445,
+      "step": 14885
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996637850391268e-06,
+      "loss": 0.4697,
+      "step": 14886
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.995517133920711e-06,
+      "loss": 0.5044,
+      "step": 14887
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994396417506463e-06,
+      "loss": 0.5122,
+      "step": 14888
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993275701162591e-06,
+      "loss": 0.447,
+      "step": 14889
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992154984903177e-06,
+      "loss": 0.4642,
+      "step": 14890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991034268742301e-06,
+      "loss": 0.4482,
+      "step": 14891
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.989913552694034e-06,
+      "loss": 0.4736,
+      "step": 14892
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98879283677245e-06,
+      "loss": 0.4944,
+      "step": 14893
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987672120991634e-06,
+      "loss": 0.491,
+      "step": 14894
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986551405365651e-06,
+      "loss": 0.4797,
+      "step": 14895
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985430689908588e-06,
+      "loss": 0.4799,
+      "step": 14896
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984309974634513e-06,
+      "loss": 0.4855,
+      "step": 14897
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983189259557508e-06,
+      "loss": 0.4503,
+      "step": 14898
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.982068544691645e-06,
+      "loss": 0.4585,
+      "step": 14899
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980947830051004e-06,
+      "loss": 0.5118,
+      "step": 14900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979827115649658e-06,
+      "loss": 0.4014,
+      "step": 14901
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978706401501684e-06,
+      "loss": 0.4208,
+      "step": 14902
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977585687621157e-06,
+      "loss": 0.4477,
+      "step": 14903
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976464974022161e-06,
+      "loss": 0.4255,
+      "step": 14904
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975344260718762e-06,
+      "loss": 0.5031,
+      "step": 14905
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97422354772504e-06,
+      "loss": 0.4838,
+      "step": 14906
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973102835055075e-06,
+      "loss": 0.4897,
+      "step": 14907
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971982122722937e-06,
+      "loss": 0.4001,
+      "step": 14908
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970861410742703e-06,
+      "loss": 0.4656,
+      "step": 14909
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969740699128458e-06,
+      "loss": 0.4517,
+      "step": 14910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968619987894266e-06,
+      "loss": 0.4871,
+      "step": 14911
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96749927705421e-06,
+      "loss": 0.4457,
+      "step": 14912
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96637856662237e-06,
+      "loss": 0.4381,
+      "step": 14913
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965257856612809e-06,
+      "loss": 0.4862,
+      "step": 14914
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.964137147039613e-06,
+      "loss": 0.4244,
+      "step": 14915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96301643791686e-06,
+      "loss": 0.4359,
+      "step": 14916
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96189572925862e-06,
+      "loss": 0.403,
+      "step": 14917
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960775021078972e-06,
+      "loss": 0.4512,
+      "step": 14918
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.959654313391995e-06,
+      "loss": 0.4695,
+      "step": 14919
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958533606211758e-06,
+      "loss": 0.5018,
+      "step": 14920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957412899552341e-06,
+      "loss": 0.4471,
+      "step": 14921
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956292193427825e-06,
+      "loss": 0.4631,
+      "step": 14922
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955171487852278e-06,
+      "loss": 0.4497,
+      "step": 14923
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95405078283978e-06,
+      "loss": 0.4484,
+      "step": 14924
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952930078404408e-06,
+      "loss": 0.4788,
+      "step": 14925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951809374560236e-06,
+      "loss": 0.4723,
+      "step": 14926
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950688671321338e-06,
+      "loss": 0.4756,
+      "step": 14927
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949567968701798e-06,
+      "loss": 0.4336,
+      "step": 14928
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.948447266715686e-06,
+      "loss": 0.4588,
+      "step": 14929
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947326565377077e-06,
+      "loss": 0.4833,
+      "step": 14930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946205864700052e-06,
+      "loss": 0.4442,
+      "step": 14931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945085164698682e-06,
+      "loss": 0.4464,
+      "step": 14932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943964465387047e-06,
+      "loss": 0.4394,
+      "step": 14933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94284376677922e-06,
+      "loss": 0.3869,
+      "step": 14934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941723068889279e-06,
+      "loss": 0.4696,
+      "step": 14935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9406023717313e-06,
+      "loss": 0.4479,
+      "step": 14936
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93948167531936e-06,
+      "loss": 0.4465,
+      "step": 14937
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93836097966753e-06,
+      "loss": 0.4358,
+      "step": 14938
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937240284789892e-06,
+      "loss": 0.4617,
+      "step": 14939
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936119590700516e-06,
+      "loss": 0.485,
+      "step": 14940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934998897413488e-06,
+      "loss": 0.4352,
+      "step": 14941
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933878204942872e-06,
+      "loss": 0.4484,
+      "step": 14942
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93275751330275e-06,
+      "loss": 0.4342,
+      "step": 14943
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9316368225072e-06,
+      "loss": 0.4209,
+      "step": 14944
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930516132570295e-06,
+      "loss": 0.51,
+      "step": 14945
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929395443506109e-06,
+      "loss": 0.4406,
+      "step": 14946
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928274755328724e-06,
+      "loss": 0.4588,
+      "step": 14947
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927154068052207e-06,
+      "loss": 0.4792,
+      "step": 14948
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926033381690642e-06,
+      "loss": 0.5014,
+      "step": 14949
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.924912696258104e-06,
+      "loss": 0.4267,
+      "step": 14950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923792011768662e-06,
+      "loss": 0.4767,
+      "step": 14951
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.922671328236398e-06,
+      "loss": 0.4101,
+      "step": 14952
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.921550645675389e-06,
+      "loss": 0.4814,
+      "step": 14953
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920429964099707e-06,
+      "loss": 0.4643,
+      "step": 14954
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919309283523426e-06,
+      "loss": 0.4893,
+      "step": 14955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918188603960632e-06,
+      "loss": 0.5041,
+      "step": 14956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917067925425387e-06,
+      "loss": 0.4146,
+      "step": 14957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915947247931775e-06,
+      "loss": 0.431,
+      "step": 14958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.914826571493873e-06,
+      "loss": 0.3809,
+      "step": 14959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.913705896125752e-06,
+      "loss": 0.4367,
+      "step": 14960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91258522184149e-06,
+      "loss": 0.4428,
+      "step": 14961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.911464548655163e-06,
+      "loss": 0.4423,
+      "step": 14962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910343876580845e-06,
+      "loss": 0.4733,
+      "step": 14963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909223205632613e-06,
+      "loss": 0.4523,
+      "step": 14964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908102535824545e-06,
+      "loss": 0.4623,
+      "step": 14965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906981867170712e-06,
+      "loss": 0.4623,
+      "step": 14966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.905861199685191e-06,
+      "loss": 0.4816,
+      "step": 14967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904740533382062e-06,
+      "loss": 0.4764,
+      "step": 14968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.903619868275393e-06,
+      "loss": 0.4992,
+      "step": 14969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902499204379266e-06,
+      "loss": 0.4414,
+      "step": 14970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901378541707758e-06,
+      "loss": 0.4439,
+      "step": 14971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.900257880274936e-06,
+      "loss": 0.4706,
+      "step": 14972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.89913722009488e-06,
+      "loss": 0.4729,
+      "step": 14973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898016561181672e-06,
+      "loss": 0.4752,
+      "step": 14974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896895903549376e-06,
+      "loss": 0.4621,
+      "step": 14975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895775247212076e-06,
+      "loss": 0.4578,
+      "step": 14976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894654592183842e-06,
+      "loss": 0.4895,
+      "step": 14977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893533938478756e-06,
+      "loss": 0.4428,
+      "step": 14978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.892413286110886e-06,
+      "loss": 0.4459,
+      "step": 14979
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891292635094314e-06,
+      "loss": 0.4979,
+      "step": 14980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.890171985443114e-06,
+      "loss": 0.4434,
+      "step": 14981
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.889051337171355e-06,
+      "loss": 0.4309,
+      "step": 14982
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88793069029312e-06,
+      "loss": 0.4368,
+      "step": 14983
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886810044822484e-06,
+      "loss": 0.4721,
+      "step": 14984
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885689400773516e-06,
+      "loss": 0.4755,
+      "step": 14985
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884568758160298e-06,
+      "loss": 0.4519,
+      "step": 14986
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883448116996906e-06,
+      "loss": 0.3738,
+      "step": 14987
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.882327477297407e-06,
+      "loss": 0.4657,
+      "step": 14988
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88120683907588e-06,
+      "loss": 0.4684,
+      "step": 14989
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880086202346409e-06,
+      "loss": 0.4399,
+      "step": 14990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878965567123059e-06,
+      "loss": 0.4284,
+      "step": 14991
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877844933419904e-06,
+      "loss": 0.47,
+      "step": 14992
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876724301251029e-06,
+      "loss": 0.4467,
+      "step": 14993
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8756036706305e-06,
+      "loss": 0.4498,
+      "step": 14994
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874483041572394e-06,
+      "loss": 0.4419,
+      "step": 14995
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873362414090795e-06,
+      "loss": 0.4413,
+      "step": 14996
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872241788199767e-06,
+      "loss": 0.4385,
+      "step": 14997
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871121163913388e-06,
+      "loss": 0.4414,
+      "step": 14998
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87000054124574e-06,
+      "loss": 0.4046,
+      "step": 14999
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868879920210887e-06,
+      "loss": 0.4235,
+      "step": 15000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.867759300822911e-06,
+      "loss": 0.4488,
+      "step": 15001
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866638683095885e-06,
+      "loss": 0.4506,
+      "step": 15002
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865518067043886e-06,
+      "loss": 0.4749,
+      "step": 15003
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864397452680985e-06,
+      "loss": 0.4642,
+      "step": 15004
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863276840021265e-06,
+      "loss": 0.4611,
+      "step": 15005
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86215622907879e-06,
+      "loss": 0.4381,
+      "step": 15006
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.861035619867642e-06,
+      "loss": 0.4823,
+      "step": 15007
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859915012401899e-06,
+      "loss": 0.4997,
+      "step": 15008
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858794406695626e-06,
+      "loss": 0.4904,
+      "step": 15009
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857673802762902e-06,
+      "loss": 0.4704,
+      "step": 15010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856553200617805e-06,
+      "loss": 0.4532,
+      "step": 15011
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.85543260027441e-06,
+      "loss": 0.4529,
+      "step": 15012
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.854312001746788e-06,
+      "loss": 0.4279,
+      "step": 15013
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853191405049014e-06,
+      "loss": 0.5058,
+      "step": 15014
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852070810195167e-06,
+      "loss": 0.4985,
+      "step": 15015
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850950217199315e-06,
+      "loss": 0.4777,
+      "step": 15016
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84982962607554e-06,
+      "loss": 0.4679,
+      "step": 15017
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848709036837914e-06,
+      "loss": 0.4378,
+      "step": 15018
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847588449500508e-06,
+      "loss": 0.4777,
+      "step": 15019
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8464678640774e-06,
+      "loss": 0.474,
+      "step": 15020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.845347280582666e-06,
+      "loss": 0.4761,
+      "step": 15021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844226699030376e-06,
+      "loss": 0.4511,
+      "step": 15022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84310611943461e-06,
+      "loss": 0.466,
+      "step": 15023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841985541809442e-06,
+      "loss": 0.4194,
+      "step": 15024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84086496616894e-06,
+      "loss": 0.4449,
+      "step": 15025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839744392527182e-06,
+      "loss": 0.4567,
+      "step": 15026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83862382089825e-06,
+      "loss": 0.4532,
+      "step": 15027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837503251296209e-06,
+      "loss": 0.4252,
+      "step": 15028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836382683735133e-06,
+      "loss": 0.4931,
+      "step": 15029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.835262118229104e-06,
+      "loss": 0.4356,
+      "step": 15030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.834141554792188e-06,
+      "loss": 0.5006,
+      "step": 15031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833020993438464e-06,
+      "loss": 0.4857,
+      "step": 15032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.831900434182009e-06,
+      "loss": 0.4524,
+      "step": 15033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830779877036892e-06,
+      "loss": 0.4645,
+      "step": 15034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829659322017188e-06,
+      "loss": 0.4749,
+      "step": 15035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828538769136977e-06,
+      "loss": 0.4526,
+      "step": 15036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827418218410321e-06,
+      "loss": 0.4389,
+      "step": 15037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826297669851306e-06,
+      "loss": 0.4029,
+      "step": 15038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.825177123474003e-06,
+      "loss": 0.4434,
+      "step": 15039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824056579292485e-06,
+      "loss": 0.456,
+      "step": 15040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822936037320824e-06,
+      "loss": 0.5037,
+      "step": 15041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8218154975731e-06,
+      "loss": 0.4552,
+      "step": 15042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82069496006338e-06,
+      "loss": 0.4518,
+      "step": 15043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819574424805743e-06,
+      "loss": 0.4148,
+      "step": 15044
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818453891814262e-06,
+      "loss": 0.4027,
+      "step": 15045
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817333361103007e-06,
+      "loss": 0.4784,
+      "step": 15046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.816212832686054e-06,
+      "loss": 0.4876,
+      "step": 15047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.815092306577483e-06,
+      "loss": 0.5039,
+      "step": 15048
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813971782791362e-06,
+      "loss": 0.5039,
+      "step": 15049
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812851261341765e-06,
+      "loss": 0.496,
+      "step": 15050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811730742242766e-06,
+      "loss": 0.4807,
+      "step": 15051
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81061022550844e-06,
+      "loss": 0.4837,
+      "step": 15052
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80948971115286e-06,
+      "loss": 0.4423,
+      "step": 15053
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8083691991901e-06,
+      "loss": 0.4911,
+      "step": 15054
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807248689634237e-06,
+      "loss": 0.437,
+      "step": 15055
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.806128182499336e-06,
+      "loss": 0.4412,
+      "step": 15056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805007677799477e-06,
+      "loss": 0.5216,
+      "step": 15057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803887175548736e-06,
+      "loss": 0.4793,
+      "step": 15058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802766675761178e-06,
+      "loss": 0.4403,
+      "step": 15059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801646178450884e-06,
+      "loss": 0.3936,
+      "step": 15060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800525683631929e-06,
+      "loss": 0.4254,
+      "step": 15061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.799405191318376e-06,
+      "loss": 0.3929,
+      "step": 15062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798284701524307e-06,
+      "loss": 0.4274,
+      "step": 15063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797164214263796e-06,
+      "loss": 0.4896,
+      "step": 15064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796043729550912e-06,
+      "loss": 0.4582,
+      "step": 15065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79492324739973e-06,
+      "loss": 0.4666,
+      "step": 15066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793802767824325e-06,
+      "loss": 0.5109,
+      "step": 15067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792682290838767e-06,
+      "loss": 0.443,
+      "step": 15068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79156181645713e-06,
+      "loss": 0.4773,
+      "step": 15069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79044134469349e-06,
+      "loss": 0.4596,
+      "step": 15070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78932087556192e-06,
+      "loss": 0.4955,
+      "step": 15071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788200409076489e-06,
+      "loss": 0.4819,
+      "step": 15072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787079945251275e-06,
+      "loss": 0.4957,
+      "step": 15073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.785959484100346e-06,
+      "loss": 0.4489,
+      "step": 15074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78483902563778e-06,
+      "loss": 0.4267,
+      "step": 15075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.783718569877646e-06,
+      "loss": 0.4585,
+      "step": 15076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78259811683402e-06,
+      "loss": 0.5394,
+      "step": 15077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.781477666520973e-06,
+      "loss": 0.4468,
+      "step": 15078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.780357218952583e-06,
+      "loss": 0.4563,
+      "step": 15079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.779236774142912e-06,
+      "loss": 0.5074,
+      "step": 15080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.778116332106044e-06,
+      "loss": 0.4034,
+      "step": 15081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77699589285605e-06,
+      "loss": 0.4213,
+      "step": 15082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.775875456406994e-06,
+      "loss": 0.3936,
+      "step": 15083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774755022772955e-06,
+      "loss": 0.4818,
+      "step": 15084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773634591968007e-06,
+      "loss": 0.4621,
+      "step": 15085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772514164006224e-06,
+      "loss": 0.4755,
+      "step": 15086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.771393738901672e-06,
+      "loss": 0.4952,
+      "step": 15087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770273316668427e-06,
+      "loss": 0.4044,
+      "step": 15088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769152897320567e-06,
+      "loss": 0.4369,
+      "step": 15089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.768032480872156e-06,
+      "loss": 0.4629,
+      "step": 15090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76691206733727e-06,
+      "loss": 0.4994,
+      "step": 15091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765791656729983e-06,
+      "loss": 0.3927,
+      "step": 15092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764671249064364e-06,
+      "loss": 0.4731,
+      "step": 15093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763550844354488e-06,
+      "loss": 0.4838,
+      "step": 15094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762430442614428e-06,
+      "loss": 0.4346,
+      "step": 15095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.761310043858255e-06,
+      "loss": 0.4403,
+      "step": 15096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760189648100036e-06,
+      "loss": 0.453,
+      "step": 15097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759069255353857e-06,
+      "loss": 0.47,
+      "step": 15098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757948865633775e-06,
+      "loss": 0.4697,
+      "step": 15099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75682847895387e-06,
+      "loss": 0.4445,
+      "step": 15100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755708095328214e-06,
+      "loss": 0.5234,
+      "step": 15101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75458771477088e-06,
+      "loss": 0.486,
+      "step": 15102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.753467337295933e-06,
+      "loss": 0.406,
+      "step": 15103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752346962917457e-06,
+      "loss": 0.4672,
+      "step": 15104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75122659164951e-06,
+      "loss": 0.4772,
+      "step": 15105
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.750106223506174e-06,
+      "loss": 0.5004,
+      "step": 15106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74898585850152e-06,
+      "loss": 0.4388,
+      "step": 15107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747865496649615e-06,
+      "loss": 0.4492,
+      "step": 15108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.746745137964532e-06,
+      "loss": 0.406,
+      "step": 15109
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.745624782460347e-06,
+      "loss": 0.4396,
+      "step": 15110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744504430151127e-06,
+      "loss": 0.3969,
+      "step": 15111
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.743384081050946e-06,
+      "loss": 0.4601,
+      "step": 15112
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74226373517388e-06,
+      "loss": 0.4127,
+      "step": 15113
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74114339253399e-06,
+      "loss": 0.4506,
+      "step": 15114
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.740023053145354e-06,
+      "loss": 0.4773,
+      "step": 15115
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.738902717022048e-06,
+      "loss": 0.4955,
+      "step": 15116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737782384178133e-06,
+      "loss": 0.3958,
+      "step": 15117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.736662054627688e-06,
+      "loss": 0.4009,
+      "step": 15118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.735541728384784e-06,
+      "loss": 0.4098,
+      "step": 15119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734421405463488e-06,
+      "loss": 0.4464,
+      "step": 15120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733301085877873e-06,
+      "loss": 0.4312,
+      "step": 15121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.732180769642013e-06,
+      "loss": 0.4327,
+      "step": 15122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73106045676998e-06,
+      "loss": 0.4872,
+      "step": 15123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72994014727584e-06,
+      "loss": 0.4447,
+      "step": 15124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728819841173666e-06,
+      "loss": 0.4556,
+      "step": 15125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727699538477534e-06,
+      "loss": 0.482,
+      "step": 15126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726579239201507e-06,
+      "loss": 0.4268,
+      "step": 15127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.725458943359662e-06,
+      "loss": 0.5145,
+      "step": 15128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72433865096607e-06,
+      "loss": 0.456,
+      "step": 15129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723218362034795e-06,
+      "loss": 0.4751,
+      "step": 15130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.722098076579912e-06,
+      "loss": 0.4068,
+      "step": 15131
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7209777946155e-06,
+      "loss": 0.3956,
+      "step": 15132
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.719857516155619e-06,
+      "loss": 0.4375,
+      "step": 15133
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718737241214341e-06,
+      "loss": 0.4594,
+      "step": 15134
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717616969805745e-06,
+      "loss": 0.4378,
+      "step": 15135
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71649670194389e-06,
+      "loss": 0.499,
+      "step": 15136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715376437642852e-06,
+      "loss": 0.4473,
+      "step": 15137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.714256176916706e-06,
+      "loss": 0.4799,
+      "step": 15138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.713135919779516e-06,
+      "loss": 0.4273,
+      "step": 15139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712015666245354e-06,
+      "loss": 0.3952,
+      "step": 15140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.710895416328295e-06,
+      "loss": 0.4571,
+      "step": 15141
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709775170042401e-06,
+      "loss": 0.447,
+      "step": 15142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70865492740175e-06,
+      "loss": 0.4577,
+      "step": 15143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70753468842041e-06,
+      "loss": 0.429,
+      "step": 15144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70641445311245e-06,
+      "loss": 0.4402,
+      "step": 15145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.705294221491938e-06,
+      "loss": 0.4628,
+      "step": 15146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704173993572951e-06,
+      "loss": 0.463,
+      "step": 15147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703053769369554e-06,
+      "loss": 0.4572,
+      "step": 15148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701933548895818e-06,
+      "loss": 0.4717,
+      "step": 15149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700813332165816e-06,
+      "loss": 0.4362,
+      "step": 15150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.699693119193611e-06,
+      "loss": 0.4601,
+      "step": 15151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698572909993276e-06,
+      "loss": 0.4701,
+      "step": 15152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697452704578887e-06,
+      "loss": 0.4454,
+      "step": 15153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.696332502964506e-06,
+      "loss": 0.469,
+      "step": 15154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.695212305164208e-06,
+      "loss": 0.4422,
+      "step": 15155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694092111192062e-06,
+      "loss": 0.5089,
+      "step": 15156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69297192106213e-06,
+      "loss": 0.4688,
+      "step": 15157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691851734788492e-06,
+      "loss": 0.5049,
+      "step": 15158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690731552385211e-06,
+      "loss": 0.4517,
+      "step": 15159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689611373866365e-06,
+      "loss": 0.4495,
+      "step": 15160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.688491199246011e-06,
+      "loss": 0.3924,
+      "step": 15161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687371028538226e-06,
+      "loss": 0.4119,
+      "step": 15162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686250861757081e-06,
+      "loss": 0.4212,
+      "step": 15163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685130698916641e-06,
+      "loss": 0.3974,
+      "step": 15164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.684010540030978e-06,
+      "loss": 0.494,
+      "step": 15165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682890385114162e-06,
+      "loss": 0.4745,
+      "step": 15166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681770234180257e-06,
+      "loss": 0.5095,
+      "step": 15167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680650087243335e-06,
+      "loss": 0.4352,
+      "step": 15168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.67952994431747e-06,
+      "loss": 0.4464,
+      "step": 15169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.678409805416726e-06,
+      "loss": 0.4355,
+      "step": 15170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.67728967055517e-06,
+      "loss": 0.4576,
+      "step": 15171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.676169539746878e-06,
+      "loss": 0.4532,
+      "step": 15172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67504941300591e-06,
+      "loss": 0.4745,
+      "step": 15173
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67392929034634e-06,
+      "loss": 0.4426,
+      "step": 15174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67280917178224e-06,
+      "loss": 0.461,
+      "step": 15175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.671689057327674e-06,
+      "loss": 0.5222,
+      "step": 15176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670568946996709e-06,
+      "loss": 0.419,
+      "step": 15177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.669448840803422e-06,
+      "loss": 0.4223,
+      "step": 15178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668328738761869e-06,
+      "loss": 0.3978,
+      "step": 15179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66720864088613e-06,
+      "loss": 0.4498,
+      "step": 15180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66608854719027e-06,
+      "loss": 0.4874,
+      "step": 15181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664968457688355e-06,
+      "loss": 0.4186,
+      "step": 15182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.663848372394452e-06,
+      "loss": 0.4483,
+      "step": 15183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.662728291322638e-06,
+      "loss": 0.4482,
+      "step": 15184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66160821448697e-06,
+      "loss": 0.3979,
+      "step": 15185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660488141901526e-06,
+      "loss": 0.4418,
+      "step": 15186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65936807358037e-06,
+      "loss": 0.4783,
+      "step": 15187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658248009537567e-06,
+      "loss": 0.4906,
+      "step": 15188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.657127949787188e-06,
+      "loss": 0.4375,
+      "step": 15189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.656007894343306e-06,
+      "loss": 0.4344,
+      "step": 15190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654887843219978e-06,
+      "loss": 0.4686,
+      "step": 15191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65376779643128e-06,
+      "loss": 0.4697,
+      "step": 15192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652647753991284e-06,
+      "loss": 0.4423,
+      "step": 15193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651527715914042e-06,
+      "loss": 0.4498,
+      "step": 15194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.650407682213635e-06,
+      "loss": 0.4407,
+      "step": 15195
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64928765290413e-06,
+      "loss": 0.4767,
+      "step": 15196
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64816762799959e-06,
+      "loss": 0.461,
+      "step": 15197
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647047607514083e-06,
+      "loss": 0.5092,
+      "step": 15198
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645927591461676e-06,
+      "loss": 0.4404,
+      "step": 15199
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644807579856445e-06,
+      "loss": 0.4793,
+      "step": 15200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.643687572712444e-06,
+      "loss": 0.427,
+      "step": 15201
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64256757004375e-06,
+      "loss": 0.4345,
+      "step": 15202
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.641447571864429e-06,
+      "loss": 0.4617,
+      "step": 15203
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640327578188544e-06,
+      "loss": 0.4012,
+      "step": 15204
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.639207589030162e-06,
+      "loss": 0.4381,
+      "step": 15205
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63808760440336e-06,
+      "loss": 0.41,
+      "step": 15206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636967624322194e-06,
+      "loss": 0.4165,
+      "step": 15207
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.635847648800734e-06,
+      "loss": 0.4597,
+      "step": 15208
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634727677853053e-06,
+      "loss": 0.4302,
+      "step": 15209
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633607711493206e-06,
+      "loss": 0.4294,
+      "step": 15210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.632487749735268e-06,
+      "loss": 0.4344,
+      "step": 15211
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631367792593308e-06,
+      "loss": 0.4719,
+      "step": 15212
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.630247840081388e-06,
+      "loss": 0.4614,
+      "step": 15213
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629127892213574e-06,
+      "loss": 0.4317,
+      "step": 15214
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628007949003936e-06,
+      "loss": 0.4639,
+      "step": 15215
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626888010466539e-06,
+      "loss": 0.4277,
+      "step": 15216
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.625768076615447e-06,
+      "loss": 0.4324,
+      "step": 15217
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624648147464735e-06,
+      "loss": 0.5075,
+      "step": 15218
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.623528223028458e-06,
+      "loss": 0.4484,
+      "step": 15219
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622408303320686e-06,
+      "loss": 0.4659,
+      "step": 15220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621288388355493e-06,
+      "loss": 0.4827,
+      "step": 15221
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620168478146935e-06,
+      "loss": 0.469,
+      "step": 15222
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619048572709083e-06,
+      "loss": 0.4619,
+      "step": 15223
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617928672056005e-06,
+      "loss": 0.4743,
+      "step": 15224
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616808776201758e-06,
+      "loss": 0.4567,
+      "step": 15225
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615688885160417e-06,
+      "loss": 0.4025,
+      "step": 15226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614568998946048e-06,
+      "loss": 0.463,
+      "step": 15227
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.613449117572712e-06,
+      "loss": 0.444,
+      "step": 15228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.612329241054474e-06,
+      "loss": 0.476,
+      "step": 15229
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611209369405404e-06,
+      "loss": 0.4197,
+      "step": 15230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610089502639569e-06,
+      "loss": 0.4946,
+      "step": 15231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608969640771027e-06,
+      "loss": 0.5233,
+      "step": 15232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.607849783813853e-06,
+      "loss": 0.4678,
+      "step": 15233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.606729931782107e-06,
+      "loss": 0.4641,
+      "step": 15234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605610084689853e-06,
+      "loss": 0.417,
+      "step": 15235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604490242551156e-06,
+      "loss": 0.451,
+      "step": 15236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60337040538009e-06,
+      "loss": 0.4541,
+      "step": 15237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602250573190712e-06,
+      "loss": 0.4858,
+      "step": 15238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601130745997084e-06,
+      "loss": 0.4472,
+      "step": 15239
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600010923813284e-06,
+      "loss": 0.4287,
+      "step": 15240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.598891106653363e-06,
+      "loss": 0.4345,
+      "step": 15241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597771294531393e-06,
+      "loss": 0.4746,
+      "step": 15242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596651487461442e-06,
+      "loss": 0.4419,
+      "step": 15243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.595531685457568e-06,
+      "loss": 0.4485,
+      "step": 15244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594411888533837e-06,
+      "loss": 0.504,
+      "step": 15245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59329209670432e-06,
+      "loss": 0.4425,
+      "step": 15246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592172309983074e-06,
+      "loss": 0.4415,
+      "step": 15247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591052528384168e-06,
+      "loss": 0.4589,
+      "step": 15248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.589932751921663e-06,
+      "loss": 0.4997,
+      "step": 15249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588812980609628e-06,
+      "loss": 0.4588,
+      "step": 15250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58769321446212e-06,
+      "loss": 0.453,
+      "step": 15251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586573453493217e-06,
+      "loss": 0.4242,
+      "step": 15252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.585453697716968e-06,
+      "loss": 0.4614,
+      "step": 15253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584333947147445e-06,
+      "loss": 0.4777,
+      "step": 15254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.583214201798713e-06,
+      "loss": 0.4514,
+      "step": 15255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582094461684831e-06,
+      "loss": 0.4432,
+      "step": 15256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.580974726819865e-06,
+      "loss": 0.4303,
+      "step": 15257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.579854997217886e-06,
+      "loss": 0.4152,
+      "step": 15258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578735272892946e-06,
+      "loss": 0.4634,
+      "step": 15259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.577615553859117e-06,
+      "loss": 0.4731,
+      "step": 15260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576495840130462e-06,
+      "loss": 0.4297,
+      "step": 15261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575376131721039e-06,
+      "loss": 0.4437,
+      "step": 15262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574256428644916e-06,
+      "loss": 0.4293,
+      "step": 15263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57313673091616e-06,
+      "loss": 0.4601,
+      "step": 15264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57201703854883e-06,
+      "loss": 0.4873,
+      "step": 15265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.570897351556985e-06,
+      "loss": 0.465,
+      "step": 15266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.569777669954694e-06,
+      "loss": 0.4533,
+      "step": 15267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.568657993756026e-06,
+      "loss": 0.4883,
+      "step": 15268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56753832297503e-06,
+      "loss": 0.4575,
+      "step": 15269
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566418657625782e-06,
+      "loss": 0.4851,
+      "step": 15270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56529899772234e-06,
+      "loss": 0.4811,
+      "step": 15271
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.564179343278764e-06,
+      "loss": 0.4244,
+      "step": 15272
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56305969430912e-06,
+      "loss": 0.4798,
+      "step": 15273
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.561940050827474e-06,
+      "loss": 0.4879,
+      "step": 15274
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.560820412847885e-06,
+      "loss": 0.4122,
+      "step": 15275
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.559700780384412e-06,
+      "loss": 0.5281,
+      "step": 15276
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.558581153451128e-06,
+      "loss": 0.4849,
+      "step": 15277
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557461532062083e-06,
+      "loss": 0.4148,
+      "step": 15278
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.556341916231347e-06,
+      "loss": 0.4125,
+      "step": 15279
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555222305972986e-06,
+      "loss": 0.4792,
+      "step": 15280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554102701301055e-06,
+      "loss": 0.436,
+      "step": 15281
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552983102229618e-06,
+      "loss": 0.448,
+      "step": 15282
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551863508772742e-06,
+      "loss": 0.4654,
+      "step": 15283
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55074392094448e-06,
+      "loss": 0.4673,
+      "step": 15284
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549624338758904e-06,
+      "loss": 0.4479,
+      "step": 15285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54850476223007e-06,
+      "loss": 0.4689,
+      "step": 15286
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547385191372041e-06,
+      "loss": 0.4493,
+      "step": 15287
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.546265626198879e-06,
+      "loss": 0.494,
+      "step": 15288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545146066724648e-06,
+      "loss": 0.4691,
+      "step": 15289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544026512963405e-06,
+      "loss": 0.4196,
+      "step": 15290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542906964929218e-06,
+      "loss": 0.4846,
+      "step": 15291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541787422636146e-06,
+      "loss": 0.4899,
+      "step": 15292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540667886098245e-06,
+      "loss": 0.4326,
+      "step": 15293
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.539548355329581e-06,
+      "loss": 0.454,
+      "step": 15294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53842883034422e-06,
+      "loss": 0.4728,
+      "step": 15295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537309311156214e-06,
+      "loss": 0.4714,
+      "step": 15296
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536189797779632e-06,
+      "loss": 0.512,
+      "step": 15297
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.535070290228534e-06,
+      "loss": 0.439,
+      "step": 15298
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533950788516974e-06,
+      "loss": 0.481,
+      "step": 15299
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532831292659019e-06,
+      "loss": 0.4465,
+      "step": 15300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531711802668733e-06,
+      "loss": 0.4354,
+      "step": 15301
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530592318560172e-06,
+      "loss": 0.4736,
+      "step": 15302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.529472840347395e-06,
+      "loss": 0.5478,
+      "step": 15303
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.528353368044465e-06,
+      "loss": 0.47,
+      "step": 15304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527233901665447e-06,
+      "loss": 0.4796,
+      "step": 15305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526114441224394e-06,
+      "loss": 0.4105,
+      "step": 15306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524994986735373e-06,
+      "loss": 0.5004,
+      "step": 15307
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523875538212443e-06,
+      "loss": 0.4408,
+      "step": 15308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.522756095669659e-06,
+      "loss": 0.459,
+      "step": 15309
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521636659121086e-06,
+      "loss": 0.4716,
+      "step": 15310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.520517228580785e-06,
+      "loss": 0.4772,
+      "step": 15311
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519397804062814e-06,
+      "loss": 0.4265,
+      "step": 15312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51827838558123e-06,
+      "loss": 0.4903,
+      "step": 15313
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517158973150103e-06,
+      "loss": 0.4525,
+      "step": 15314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51603956678348e-06,
+      "loss": 0.4381,
+      "step": 15315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51492016649543e-06,
+      "loss": 0.4499,
+      "step": 15316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513800772300012e-06,
+      "loss": 0.4601,
+      "step": 15317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.512681384211281e-06,
+      "loss": 0.486,
+      "step": 15318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511562002243297e-06,
+      "loss": 0.4302,
+      "step": 15319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510442626410127e-06,
+      "loss": 0.4321,
+      "step": 15320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50932325672582e-06,
+      "loss": 0.4408,
+      "step": 15321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508203893204444e-06,
+      "loss": 0.4296,
+      "step": 15322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507084535860053e-06,
+      "loss": 0.456,
+      "step": 15323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.505965184706708e-06,
+      "loss": 0.4586,
+      "step": 15324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504845839758466e-06,
+      "loss": 0.4413,
+      "step": 15325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503726501029393e-06,
+      "loss": 0.4424,
+      "step": 15326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.502607168533538e-06,
+      "loss": 0.4343,
+      "step": 15327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501487842284967e-06,
+      "loss": 0.4545,
+      "step": 15328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500368522297738e-06,
+      "loss": 0.4512,
+      "step": 15329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499249208585905e-06,
+      "loss": 0.4463,
+      "step": 15330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49812990116353e-06,
+      "loss": 0.4469,
+      "step": 15331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.497010600044675e-06,
+      "loss": 0.4564,
+      "step": 15332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.495891305243392e-06,
+      "loss": 0.4183,
+      "step": 15333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494772016773744e-06,
+      "loss": 0.4322,
+      "step": 15334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49365273464979e-06,
+      "loss": 0.4571,
+      "step": 15335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49253345888558e-06,
+      "loss": 0.4292,
+      "step": 15336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491414189495182e-06,
+      "loss": 0.4075,
+      "step": 15337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.490294926492652e-06,
+      "loss": 0.4549,
+      "step": 15338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489175669892045e-06,
+      "loss": 0.4252,
+      "step": 15339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488056419707417e-06,
+      "loss": 0.454,
+      "step": 15340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48693717595283e-06,
+      "loss": 0.4522,
+      "step": 15341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485817938642344e-06,
+      "loss": 0.4325,
+      "step": 15342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48469870779001e-06,
+      "loss": 0.4757,
+      "step": 15343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48357948340989e-06,
+      "loss": 0.5289,
+      "step": 15344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.482460265516045e-06,
+      "loss": 0.4448,
+      "step": 15345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481341054122521e-06,
+      "loss": 0.4403,
+      "step": 15346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480221849243385e-06,
+      "loss": 0.4625,
+      "step": 15347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479102650892696e-06,
+      "loss": 0.4571,
+      "step": 15348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477983459084502e-06,
+      "loss": 0.4737,
+      "step": 15349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.476864273832866e-06,
+      "loss": 0.4473,
+      "step": 15350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475745095151845e-06,
+      "loss": 0.4143,
+      "step": 15351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.474625923055493e-06,
+      "loss": 0.4061,
+      "step": 15352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.473506757557868e-06,
+      "loss": 0.4167,
+      "step": 15353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.472387598673033e-06,
+      "loss": 0.4673,
+      "step": 15354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471268446415036e-06,
+      "loss": 0.4804,
+      "step": 15355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470149300797935e-06,
+      "loss": 0.4299,
+      "step": 15356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469030161835794e-06,
+      "loss": 0.4446,
+      "step": 15357
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467911029542657e-06,
+      "loss": 0.4224,
+      "step": 15358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.466791903932591e-06,
+      "loss": 0.4595,
+      "step": 15359
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465672785019651e-06,
+      "loss": 0.4616,
+      "step": 15360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.464553672817886e-06,
+      "loss": 0.4467,
+      "step": 15361
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.463434567341358e-06,
+      "loss": 0.4219,
+      "step": 15362
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.462315468604126e-06,
+      "loss": 0.4734,
+      "step": 15363
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.461196376620236e-06,
+      "loss": 0.5009,
+      "step": 15364
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.460077291403752e-06,
+      "loss": 0.4091,
+      "step": 15365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45895821296873e-06,
+      "loss": 0.465,
+      "step": 15366
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457839141329219e-06,
+      "loss": 0.4407,
+      "step": 15367
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45672007649928e-06,
+      "loss": 0.417,
+      "step": 15368
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45560101849297e-06,
+      "loss": 0.4497,
+      "step": 15369
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45448196732434e-06,
+      "loss": 0.5028,
+      "step": 15370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453362923007445e-06,
+      "loss": 0.4155,
+      "step": 15371
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.452243885556346e-06,
+      "loss": 0.4321,
+      "step": 15372
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45112485498509e-06,
+      "loss": 0.4355,
+      "step": 15373
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.450005831307736e-06,
+      "loss": 0.4406,
+      "step": 15374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.448886814538346e-06,
+      "loss": 0.4842,
+      "step": 15375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.447767804690966e-06,
+      "loss": 0.4412,
+      "step": 15376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44664880177965e-06,
+      "loss": 0.5006,
+      "step": 15377
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.445529805818456e-06,
+      "loss": 0.4404,
+      "step": 15378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444410816821443e-06,
+      "loss": 0.4635,
+      "step": 15379
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44329183480266e-06,
+      "loss": 0.4438,
+      "step": 15380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44217285977616e-06,
+      "loss": 0.4314,
+      "step": 15381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441053891756004e-06,
+      "loss": 0.4195,
+      "step": 15382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439934930756239e-06,
+      "loss": 0.4518,
+      "step": 15383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.438815976790923e-06,
+      "loss": 0.4248,
+      "step": 15384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437697029874113e-06,
+      "loss": 0.4641,
+      "step": 15385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436578090019858e-06,
+      "loss": 0.3829,
+      "step": 15386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.435459157242211e-06,
+      "loss": 0.5309,
+      "step": 15387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434340231555235e-06,
+      "loss": 0.4731,
+      "step": 15388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43322131297297e-06,
+      "loss": 0.4474,
+      "step": 15389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.432102401509481e-06,
+      "loss": 0.4188,
+      "step": 15390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430983497178818e-06,
+      "loss": 0.4601,
+      "step": 15391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.429864599995035e-06,
+      "loss": 0.4341,
+      "step": 15392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428745709972181e-06,
+      "loss": 0.3976,
+      "step": 15393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427626827124316e-06,
+      "loss": 0.438,
+      "step": 15394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426507951465487e-06,
+      "loss": 0.3965,
+      "step": 15395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425389083009753e-06,
+      "loss": 0.4541,
+      "step": 15396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.424270221771166e-06,
+      "loss": 0.4178,
+      "step": 15397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423151367763771e-06,
+      "loss": 0.4366,
+      "step": 15398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42203252100163e-06,
+      "loss": 0.4787,
+      "step": 15399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420913681498796e-06,
+      "loss": 0.466,
+      "step": 15400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419794849269315e-06,
+      "loss": 0.4314,
+      "step": 15401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418676024327245e-06,
+      "loss": 0.4251,
+      "step": 15402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417557206686639e-06,
+      "loss": 0.4429,
+      "step": 15403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416438396361542e-06,
+      "loss": 0.4456,
+      "step": 15404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415319593366012e-06,
+      "loss": 0.4401,
+      "step": 15405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414200797714104e-06,
+      "loss": 0.4471,
+      "step": 15406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413082009419866e-06,
+      "loss": 0.4558,
+      "step": 15407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.411963228497347e-06,
+      "loss": 0.4271,
+      "step": 15408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410844454960608e-06,
+      "loss": 0.4611,
+      "step": 15409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.409725688823692e-06,
+      "loss": 0.439,
+      "step": 15410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408606930100655e-06,
+      "loss": 0.4869,
+      "step": 15411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40748817880555e-06,
+      "loss": 0.4395,
+      "step": 15412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.406369434952427e-06,
+      "loss": 0.5064,
+      "step": 15413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405250698555334e-06,
+      "loss": 0.4727,
+      "step": 15414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404131969628325e-06,
+      "loss": 0.4634,
+      "step": 15415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.403013248185456e-06,
+      "loss": 0.4582,
+      "step": 15416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401894534240774e-06,
+      "loss": 0.4212,
+      "step": 15417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400775827808325e-06,
+      "loss": 0.5203,
+      "step": 15418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.399657128902172e-06,
+      "loss": 0.4828,
+      "step": 15419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.398538437536352e-06,
+      "loss": 0.4557,
+      "step": 15420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.397419753724924e-06,
+      "loss": 0.4298,
+      "step": 15421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.396301077481944e-06,
+      "loss": 0.4735,
+      "step": 15422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395182408821452e-06,
+      "loss": 0.4627,
+      "step": 15423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394063747757502e-06,
+      "loss": 0.4565,
+      "step": 15424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392945094304148e-06,
+      "loss": 0.4813,
+      "step": 15425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.391826448475434e-06,
+      "loss": 0.4414,
+      "step": 15426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390707810285416e-06,
+      "loss": 0.4118,
+      "step": 15427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389589179748142e-06,
+      "loss": 0.3973,
+      "step": 15428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388470556877662e-06,
+      "loss": 0.5088,
+      "step": 15429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387351941688026e-06,
+      "loss": 0.4616,
+      "step": 15430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386233334193285e-06,
+      "loss": 0.4444,
+      "step": 15431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.385114734407485e-06,
+      "loss": 0.438,
+      "step": 15432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38399614234468e-06,
+      "loss": 0.4372,
+      "step": 15433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38287755801892e-06,
+      "loss": 0.4522,
+      "step": 15434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381758981444251e-06,
+      "loss": 0.4579,
+      "step": 15435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38064041263472e-06,
+      "loss": 0.4873,
+      "step": 15436
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379521851604385e-06,
+      "loss": 0.4348,
+      "step": 15437
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378403298367288e-06,
+      "loss": 0.4666,
+      "step": 15438
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37728475293748e-06,
+      "loss": 0.3851,
+      "step": 15439
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376166215329015e-06,
+      "loss": 0.4652,
+      "step": 15440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37504768555593e-06,
+      "loss": 0.4567,
+      "step": 15441
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.373929163632286e-06,
+      "loss": 0.5003,
+      "step": 15442
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372810649572125e-06,
+      "loss": 0.507,
+      "step": 15443
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.371692143389498e-06,
+      "loss": 0.4645,
+      "step": 15444
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370573645098454e-06,
+      "loss": 0.5061,
+      "step": 15445
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36945515471304e-06,
+      "loss": 0.4656,
+      "step": 15446
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368336672247302e-06,
+      "loss": 0.4658,
+      "step": 15447
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.367218197715291e-06,
+      "loss": 0.4255,
+      "step": 15448
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366099731131055e-06,
+      "loss": 0.4841,
+      "step": 15449
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364981272508647e-06,
+      "loss": 0.3868,
+      "step": 15450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363862821862104e-06,
+      "loss": 0.4178,
+      "step": 15451
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36274437920548e-06,
+      "loss": 0.4474,
+      "step": 15452
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361625944552825e-06,
+      "loss": 0.4084,
+      "step": 15453
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360507517918184e-06,
+      "loss": 0.469,
+      "step": 15454
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359389099315601e-06,
+      "loss": 0.4265,
+      "step": 15455
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.358270688759131e-06,
+      "loss": 0.4595,
+      "step": 15456
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357152286262812e-06,
+      "loss": 0.4442,
+      "step": 15457
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356033891840697e-06,
+      "loss": 0.4584,
+      "step": 15458
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.354915505506839e-06,
+      "loss": 0.4543,
+      "step": 15459
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353797127275274e-06,
+      "loss": 0.4599,
+      "step": 15460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.35267875716005e-06,
+      "loss": 0.4774,
+      "step": 15461
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.351560395175225e-06,
+      "loss": 0.4254,
+      "step": 15462
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.350442041334831e-06,
+      "loss": 0.4743,
+      "step": 15463
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.349323695652923e-06,
+      "loss": 0.4638,
+      "step": 15464
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.348205358143547e-06,
+      "loss": 0.488,
+      "step": 15465
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347087028820747e-06,
+      "loss": 0.4496,
+      "step": 15466
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.4326,
+      "step": 15467
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.344850394791065e-06,
+      "loss": 0.4807,
+      "step": 15468
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343732090112274e-06,
+      "loss": 0.4496,
+      "step": 15469
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.342613793676245e-06,
+      "loss": 0.4839,
+      "step": 15470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341495505497027e-06,
+      "loss": 0.4471,
+      "step": 15471
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340377225588655e-06,
+      "loss": 0.4159,
+      "step": 15472
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339258953965184e-06,
+      "loss": 0.4775,
+      "step": 15473
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338140690640662e-06,
+      "loss": 0.5087,
+      "step": 15474
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337022435629127e-06,
+      "loss": 0.4626,
+      "step": 15475
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335904188944627e-06,
+      "loss": 0.4415,
+      "step": 15476
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33478595060121e-06,
+      "loss": 0.4871,
+      "step": 15477
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333667720612915e-06,
+      "loss": 0.4454,
+      "step": 15478
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332549498993792e-06,
+      "loss": 0.453,
+      "step": 15479
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.331431285757888e-06,
+      "loss": 0.4405,
+      "step": 15480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.330313080919243e-06,
+      "loss": 0.4669,
+      "step": 15481
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3291948844919e-06,
+      "loss": 0.405,
+      "step": 15482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328076696489911e-06,
+      "loss": 0.3956,
+      "step": 15483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.326958516927312e-06,
+      "loss": 0.4563,
+      "step": 15484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325840345818154e-06,
+      "loss": 0.4125,
+      "step": 15485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324722183176478e-06,
+      "loss": 0.4468,
+      "step": 15486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.323604029016333e-06,
+      "loss": 0.472,
+      "step": 15487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322485883351755e-06,
+      "loss": 0.4447,
+      "step": 15488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321367746196792e-06,
+      "loss": 0.4814,
+      "step": 15489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.320249617565493e-06,
+      "loss": 0.4649,
+      "step": 15490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319131497471894e-06,
+      "loss": 0.444,
+      "step": 15491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318013385930039e-06,
+      "loss": 0.4522,
+      "step": 15492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316895282953979e-06,
+      "loss": 0.5211,
+      "step": 15493
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.315777188557747e-06,
+      "loss": 0.451,
+      "step": 15494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314659102755393e-06,
+      "loss": 0.3923,
+      "step": 15495
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.313541025560964e-06,
+      "loss": 0.4712,
+      "step": 15496
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312422956988496e-06,
+      "loss": 0.459,
+      "step": 15497
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311304897052031e-06,
+      "loss": 0.4673,
+      "step": 15498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31018684576562e-06,
+      "loss": 0.4442,
+      "step": 15499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309068803143296e-06,
+      "loss": 0.4273,
+      "step": 15500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307950769199108e-06,
+      "loss": 0.4528,
+      "step": 15501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.306832743947101e-06,
+      "loss": 0.4792,
+      "step": 15502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.305714727401308e-06,
+      "loss": 0.4599,
+      "step": 15503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.304596719575778e-06,
+      "loss": 0.4417,
+      "step": 15504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303478720484556e-06,
+      "loss": 0.4112,
+      "step": 15505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302360730141675e-06,
+      "loss": 0.4559,
+      "step": 15506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301242748561184e-06,
+      "loss": 0.4536,
+      "step": 15507
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300124775757127e-06,
+      "loss": 0.4214,
+      "step": 15508
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299006811743537e-06,
+      "loss": 0.4711,
+      "step": 15509
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297888856534461e-06,
+      "loss": 0.4375,
+      "step": 15510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296770910143946e-06,
+      "loss": 0.4065,
+      "step": 15511
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295652972586024e-06,
+      "loss": 0.4773,
+      "step": 15512
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.294535043874738e-06,
+      "loss": 0.4466,
+      "step": 15513
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293417124024136e-06,
+      "loss": 0.46,
+      "step": 15514
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292299213048249e-06,
+      "loss": 0.46,
+      "step": 15515
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291181310961124e-06,
+      "loss": 0.4359,
+      "step": 15516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290063417776807e-06,
+      "loss": 0.4807,
+      "step": 15517
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288945533509331e-06,
+      "loss": 0.3856,
+      "step": 15518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287827658172735e-06,
+      "loss": 0.4757,
+      "step": 15519
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28670979178107e-06,
+      "loss": 0.4849,
+      "step": 15520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285591934348363e-06,
+      "loss": 0.4711,
+      "step": 15521
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284474085888666e-06,
+      "loss": 0.3843,
+      "step": 15522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28335624641601e-06,
+      "loss": 0.489,
+      "step": 15523
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282238415944446e-06,
+      "loss": 0.5245,
+      "step": 15524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.281120594488004e-06,
+      "loss": 0.4537,
+      "step": 15525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280002782060726e-06,
+      "loss": 0.4627,
+      "step": 15526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278884978676658e-06,
+      "loss": 0.4489,
+      "step": 15527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.277767184349832e-06,
+      "loss": 0.4738,
+      "step": 15528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27664939909429e-06,
+      "loss": 0.4581,
+      "step": 15529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.275531622924075e-06,
+      "loss": 0.447,
+      "step": 15530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274413855853221e-06,
+      "loss": 0.4649,
+      "step": 15531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273296097895769e-06,
+      "loss": 0.4536,
+      "step": 15532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272178349065761e-06,
+      "loss": 0.4867,
+      "step": 15533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271060609377234e-06,
+      "loss": 0.5049,
+      "step": 15534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269942878844224e-06,
+      "loss": 0.4572,
+      "step": 15535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268825157480776e-06,
+      "loss": 0.4532,
+      "step": 15536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26770744530092e-06,
+      "loss": 0.4485,
+      "step": 15537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266589742318703e-06,
+      "loss": 0.4723,
+      "step": 15538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26547204854816e-06,
+      "loss": 0.4233,
+      "step": 15539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264354364003327e-06,
+      "loss": 0.4562,
+      "step": 15540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263236688698244e-06,
+      "loss": 0.4577,
+      "step": 15541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262119022646953e-06,
+      "loss": 0.4628,
+      "step": 15542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261001365863485e-06,
+      "loss": 0.4687,
+      "step": 15543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259883718361884e-06,
+      "loss": 0.4724,
+      "step": 15544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258766080156185e-06,
+      "loss": 0.4576,
+      "step": 15545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257648451260422e-06,
+      "loss": 0.4409,
+      "step": 15546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256530831688638e-06,
+      "loss": 0.4491,
+      "step": 15547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25541322145487e-06,
+      "loss": 0.4236,
+      "step": 15548
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254295620573153e-06,
+      "loss": 0.3845,
+      "step": 15549
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253178029057522e-06,
+      "loss": 0.4451,
+      "step": 15550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252060446922021e-06,
+      "loss": 0.4518,
+      "step": 15551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25094287418068e-06,
+      "loss": 0.4906,
+      "step": 15552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249825310847535e-06,
+      "loss": 0.4126,
+      "step": 15553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.248707756936634e-06,
+      "loss": 0.417,
+      "step": 15554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.247590212462001e-06,
+      "loss": 0.4806,
+      "step": 15555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246472677437677e-06,
+      "loss": 0.4409,
+      "step": 15556
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.245355151877702e-06,
+      "loss": 0.4561,
+      "step": 15557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244237635796103e-06,
+      "loss": 0.4462,
+      "step": 15558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.243120129206925e-06,
+      "loss": 0.4685,
+      "step": 15559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.242002632124198e-06,
+      "loss": 0.4444,
+      "step": 15560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.240885144561963e-06,
+      "loss": 0.4601,
+      "step": 15561
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.239767666534252e-06,
+      "loss": 0.4558,
+      "step": 15562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.2386501980551e-06,
+      "loss": 0.4494,
+      "step": 15563
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23753273913855e-06,
+      "loss": 0.4019,
+      "step": 15564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236415289798628e-06,
+      "loss": 0.3993,
+      "step": 15565
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23529785004937e-06,
+      "loss": 0.4169,
+      "step": 15566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234180419904819e-06,
+      "loss": 0.448,
+      "step": 15567
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233062999379e-06,
+      "loss": 0.4677,
+      "step": 15568
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231945588485954e-06,
+      "loss": 0.4509,
+      "step": 15569
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230828187239716e-06,
+      "loss": 0.435,
+      "step": 15570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229710795654319e-06,
+      "loss": 0.4723,
+      "step": 15571
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228593413743795e-06,
+      "loss": 0.4082,
+      "step": 15572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227476041522186e-06,
+      "loss": 0.4531,
+      "step": 15573
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226358679003515e-06,
+      "loss": 0.4732,
+      "step": 15574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225241326201825e-06,
+      "loss": 0.3745,
+      "step": 15575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22412398313115e-06,
+      "loss": 0.454,
+      "step": 15576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223006649805518e-06,
+      "loss": 0.4664,
+      "step": 15577
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221889326238963e-06,
+      "loss": 0.4405,
+      "step": 15578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22077201244553e-06,
+      "loss": 0.4645,
+      "step": 15579
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219654708439238e-06,
+      "loss": 0.4243,
+      "step": 15580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.218537414234127e-06,
+      "loss": 0.4159,
+      "step": 15581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217420129844234e-06,
+      "loss": 0.4791,
+      "step": 15582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216302855283582e-06,
+      "loss": 0.4693,
+      "step": 15583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215185590566213e-06,
+      "loss": 0.4531,
+      "step": 15584
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214068335706158e-06,
+      "loss": 0.4533,
+      "step": 15585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21295109071745e-06,
+      "loss": 0.4308,
+      "step": 15586
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211833855614115e-06,
+      "loss": 0.4584,
+      "step": 15587
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210716630410197e-06,
+      "loss": 0.4801,
+      "step": 15588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209599415119717e-06,
+      "loss": 0.4195,
+      "step": 15589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208482209756712e-06,
+      "loss": 0.4232,
+      "step": 15590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207365014335221e-06,
+      "loss": 0.3896,
+      "step": 15591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206247828869266e-06,
+      "loss": 0.4458,
+      "step": 15592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205130653372882e-06,
+      "loss": 0.4706,
+      "step": 15593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204013487860106e-06,
+      "loss": 0.4281,
+      "step": 15594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20289633234496e-06,
+      "loss": 0.4604,
+      "step": 15595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201779186841483e-06,
+      "loss": 0.4708,
+      "step": 15596
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200662051363701e-06,
+      "loss": 0.4513,
+      "step": 15597
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199544925925655e-06,
+      "loss": 0.4121,
+      "step": 15598
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198427810541362e-06,
+      "loss": 0.4831,
+      "step": 15599
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.197310705224863e-06,
+      "loss": 0.4326,
+      "step": 15600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19619360999019e-06,
+      "loss": 0.4529,
+      "step": 15601
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195076524851369e-06,
+      "loss": 0.4351,
+      "step": 15602
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19395944982243e-06,
+      "loss": 0.4266,
+      "step": 15603
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192842384917409e-06,
+      "loss": 0.4865,
+      "step": 15604
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191725330150328e-06,
+      "loss": 0.4576,
+      "step": 15605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190608285535225e-06,
+      "loss": 0.4233,
+      "step": 15606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189491251086127e-06,
+      "loss": 0.4125,
+      "step": 15607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188374226817064e-06,
+      "loss": 0.4709,
+      "step": 15608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.187257212742064e-06,
+      "loss": 0.4985,
+      "step": 15609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186140208875163e-06,
+      "loss": 0.5179,
+      "step": 15610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185023215230383e-06,
+      "loss": 0.4652,
+      "step": 15611
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18390623182176e-06,
+      "loss": 0.4456,
+      "step": 15612
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182789258663321e-06,
+      "loss": 0.4319,
+      "step": 15613
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181672295769091e-06,
+      "loss": 0.4415,
+      "step": 15614
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180555343153103e-06,
+      "loss": 0.4054,
+      "step": 15615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17943840082939e-06,
+      "loss": 0.4365,
+      "step": 15616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178321468811971e-06,
+      "loss": 0.4646,
+      "step": 15617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177204547114884e-06,
+      "loss": 0.4359,
+      "step": 15618
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176087635752157e-06,
+      "loss": 0.4457,
+      "step": 15619
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174970734737809e-06,
+      "loss": 0.3701,
+      "step": 15620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173853844085876e-06,
+      "loss": 0.4097,
+      "step": 15621
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17273696381039e-06,
+      "loss": 0.4878,
+      "step": 15622
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171620093925371e-06,
+      "loss": 0.4876,
+      "step": 15623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170503234444847e-06,
+      "loss": 0.4435,
+      "step": 15624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169386385382853e-06,
+      "loss": 0.4682,
+      "step": 15625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16826954675341e-06,
+      "loss": 0.473,
+      "step": 15626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167152718570546e-06,
+      "loss": 0.4654,
+      "step": 15627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166035900848299e-06,
+      "loss": 0.5089,
+      "step": 15628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.164919093600681e-06,
+      "loss": 0.4403,
+      "step": 15629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163802296841725e-06,
+      "loss": 0.4523,
+      "step": 15630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162685510585464e-06,
+      "loss": 0.4007,
+      "step": 15631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.161568734845917e-06,
+      "loss": 0.4595,
+      "step": 15632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160451969637113e-06,
+      "loss": 0.4567,
+      "step": 15633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159335214973079e-06,
+      "loss": 0.4406,
+      "step": 15634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158218470867844e-06,
+      "loss": 0.4802,
+      "step": 15635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15710173733543e-06,
+      "loss": 0.4874,
+      "step": 15636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.155985014389866e-06,
+      "loss": 0.4977,
+      "step": 15637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15486830204518e-06,
+      "loss": 0.4619,
+      "step": 15638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153751600315392e-06,
+      "loss": 0.418,
+      "step": 15639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.152634909214531e-06,
+      "loss": 0.4687,
+      "step": 15640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151518228756627e-06,
+      "loss": 0.449,
+      "step": 15641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150401558955698e-06,
+      "loss": 0.4549,
+      "step": 15642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149284899825775e-06,
+      "loss": 0.4414,
+      "step": 15643
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148168251380882e-06,
+      "loss": 0.485,
+      "step": 15644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14705161363504e-06,
+      "loss": 0.4699,
+      "step": 15645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145934986602279e-06,
+      "loss": 0.4741,
+      "step": 15646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144818370296625e-06,
+      "loss": 0.4571,
+      "step": 15647
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.143701764732095e-06,
+      "loss": 0.412,
+      "step": 15648
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142585169922722e-06,
+      "loss": 0.478,
+      "step": 15649
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.141468585882532e-06,
+      "loss": 0.4779,
+      "step": 15650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140352012625538e-06,
+      "loss": 0.4243,
+      "step": 15651
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13923545016577e-06,
+      "loss": 0.4901,
+      "step": 15652
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.138118898517257e-06,
+      "loss": 0.4507,
+      "step": 15653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137002357694019e-06,
+      "loss": 0.4199,
+      "step": 15654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135885827710077e-06,
+      "loss": 0.4325,
+      "step": 15655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13476930857946e-06,
+      "loss": 0.4463,
+      "step": 15656
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133652800316186e-06,
+      "loss": 0.4102,
+      "step": 15657
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13253630293428e-06,
+      "loss": 0.4491,
+      "step": 15658
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131419816447772e-06,
+      "loss": 0.4147,
+      "step": 15659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130303340870679e-06,
+      "loss": 0.4512,
+      "step": 15660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129186876217021e-06,
+      "loss": 0.4721,
+      "step": 15661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128070422500828e-06,
+      "loss": 0.429,
+      "step": 15662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126953979736117e-06,
+      "loss": 0.4918,
+      "step": 15663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125837547936914e-06,
+      "loss": 0.4591,
+      "step": 15664
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.124721127117241e-06,
+      "loss": 0.4662,
+      "step": 15665
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12360471729112e-06,
+      "loss": 0.3734,
+      "step": 15666
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122488318472569e-06,
+      "loss": 0.4565,
+      "step": 15667
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121371930675616e-06,
+      "loss": 0.4685,
+      "step": 15668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120255553914284e-06,
+      "loss": 0.4494,
+      "step": 15669
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.119139188202588e-06,
+      "loss": 0.4361,
+      "step": 15670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118022833554552e-06,
+      "loss": 0.4302,
+      "step": 15671
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116906489984203e-06,
+      "loss": 0.4992,
+      "step": 15672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115790157505554e-06,
+      "loss": 0.4623,
+      "step": 15673
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114673836132633e-06,
+      "loss": 0.3919,
+      "step": 15674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113557525879456e-06,
+      "loss": 0.4089,
+      "step": 15675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112441226760047e-06,
+      "loss": 0.491,
+      "step": 15676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111324938788426e-06,
+      "loss": 0.4626,
+      "step": 15677
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110208661978615e-06,
+      "loss": 0.4741,
+      "step": 15678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10909239634463e-06,
+      "loss": 0.4664,
+      "step": 15679
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107976141900496e-06,
+      "loss": 0.4759,
+      "step": 15680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106859898660234e-06,
+      "loss": 0.4231,
+      "step": 15681
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105743666637858e-06,
+      "loss": 0.4875,
+      "step": 15682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10462744584739e-06,
+      "loss": 0.4823,
+      "step": 15683
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103511236302858e-06,
+      "loss": 0.4169,
+      "step": 15684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102395038018268e-06,
+      "loss": 0.4576,
+      "step": 15685
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101278851007651e-06,
+      "loss": 0.4474,
+      "step": 15686
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100162675285024e-06,
+      "loss": 0.4298,
+      "step": 15687
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0990465108644e-06,
+      "loss": 0.4705,
+      "step": 15688
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0979303577598e-06,
+      "loss": 0.4409,
+      "step": 15689
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096814215985252e-06,
+      "loss": 0.4754,
+      "step": 15690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095698085554766e-06,
+      "loss": 0.4911,
+      "step": 15691
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094581966482357e-06,
+      "loss": 0.4434,
+      "step": 15692
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093465858782056e-06,
+      "loss": 0.4148,
+      "step": 15693
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092349762467871e-06,
+      "loss": 0.4198,
+      "step": 15694
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091233677553824e-06,
+      "loss": 0.4448,
+      "step": 15695
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.090117604053936e-06,
+      "loss": 0.4185,
+      "step": 15696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08900154198222e-06,
+      "loss": 0.4769,
+      "step": 15697
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087885491352693e-06,
+      "loss": 0.4464,
+      "step": 15698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08676945217938e-06,
+      "loss": 0.4643,
+      "step": 15699
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08565342447629e-06,
+      "loss": 0.4231,
+      "step": 15700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084537408257447e-06,
+      "loss": 0.4339,
+      "step": 15701
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083421403536865e-06,
+      "loss": 0.4865,
+      "step": 15702
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08230541032856e-06,
+      "loss": 0.468,
+      "step": 15703
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08118942864655e-06,
+      "loss": 0.4378,
+      "step": 15704
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.080073458504851e-06,
+      "loss": 0.4788,
+      "step": 15705
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.078957499917485e-06,
+      "loss": 0.4612,
+      "step": 15706
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.077841552898463e-06,
+      "loss": 0.4214,
+      "step": 15707
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076725617461801e-06,
+      "loss": 0.4773,
+      "step": 15708
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075609693621521e-06,
+      "loss": 0.4369,
+      "step": 15709
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07449378139163e-06,
+      "loss": 0.4219,
+      "step": 15710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073377880786152e-06,
+      "loss": 0.4241,
+      "step": 15711
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072261991819099e-06,
+      "loss": 0.464,
+      "step": 15712
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.071146114504488e-06,
+      "loss": 0.4962,
+      "step": 15713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07003024885633e-06,
+      "loss": 0.4937,
+      "step": 15714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.068914394888651e-06,
+      "loss": 0.3876,
+      "step": 15715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067798552615453e-06,
+      "loss": 0.4609,
+      "step": 15716
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06668272205076e-06,
+      "loss": 0.4487,
+      "step": 15717
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065566903208588e-06,
+      "loss": 0.4422,
+      "step": 15718
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064451096102942e-06,
+      "loss": 0.4759,
+      "step": 15719
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063335300747843e-06,
+      "loss": 0.4381,
+      "step": 15720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06221951715731e-06,
+      "loss": 0.4337,
+      "step": 15721
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061103745345348e-06,
+      "loss": 0.5338,
+      "step": 15722
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059987985325976e-06,
+      "loss": 0.4494,
+      "step": 15723
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058872237113212e-06,
+      "loss": 0.4396,
+      "step": 15724
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05775650072106e-06,
+      "loss": 0.4518,
+      "step": 15725
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056640776163539e-06,
+      "loss": 0.4294,
+      "step": 15726
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055525063454667e-06,
+      "loss": 0.411,
+      "step": 15727
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.054409362608453e-06,
+      "loss": 0.4543,
+      "step": 15728
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053293673638906e-06,
+      "loss": 0.412,
+      "step": 15729
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052177996560048e-06,
+      "loss": 0.392,
+      "step": 15730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051062331385882e-06,
+      "loss": 0.3786,
+      "step": 15731
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049946678130427e-06,
+      "loss": 0.4543,
+      "step": 15732
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.048831036807702e-06,
+      "loss": 0.4447,
+      "step": 15733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.047715407431707e-06,
+      "loss": 0.4138,
+      "step": 15734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046599790016459e-06,
+      "loss": 0.432,
+      "step": 15735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.045484184575975e-06,
+      "loss": 0.4442,
+      "step": 15736
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044368591124258e-06,
+      "loss": 0.4191,
+      "step": 15737
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043253009675329e-06,
+      "loss": 0.486,
+      "step": 15738
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.042137440243195e-06,
+      "loss": 0.4337,
+      "step": 15739
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041021882841868e-06,
+      "loss": 0.4167,
+      "step": 15740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039906337485358e-06,
+      "loss": 0.4701,
+      "step": 15741
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.038790804187678e-06,
+      "loss": 0.4532,
+      "step": 15742
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037675282962845e-06,
+      "loss": 0.4339,
+      "step": 15743
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03655977382486e-06,
+      "loss": 0.4656,
+      "step": 15744
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035444276787736e-06,
+      "loss": 0.4774,
+      "step": 15745
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034328791865492e-06,
+      "loss": 0.4742,
+      "step": 15746
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033213319072126e-06,
+      "loss": 0.4134,
+      "step": 15747
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03209785842166e-06,
+      "loss": 0.4268,
+      "step": 15748
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.030982409928098e-06,
+      "loss": 0.4483,
+      "step": 15749
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02986697360545e-06,
+      "loss": 0.4674,
+      "step": 15750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028751549467725e-06,
+      "loss": 0.4173,
+      "step": 15751
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027636137528941e-06,
+      "loss": 0.4225,
+      "step": 15752
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.026520737803096e-06,
+      "loss": 0.4449,
+      "step": 15753
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.025405350304208e-06,
+      "loss": 0.4648,
+      "step": 15754
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024289975046284e-06,
+      "loss": 0.4839,
+      "step": 15755
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02317461204333e-06,
+      "loss": 0.4244,
+      "step": 15756
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022059261309357e-06,
+      "loss": 0.449,
+      "step": 15757
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020943922858379e-06,
+      "loss": 0.4364,
+      "step": 15758
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.4447,
+      "step": 15759
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018713282861421e-06,
+      "loss": 0.4618,
+      "step": 15760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017597981343465e-06,
+      "loss": 0.4563,
+      "step": 15761
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01648269216453e-06,
+      "loss": 0.4065,
+      "step": 15762
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015367415338628e-06,
+      "loss": 0.4395,
+      "step": 15763
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014252150879768e-06,
+      "loss": 0.456,
+      "step": 15764
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013136898801956e-06,
+      "loss": 0.4837,
+      "step": 15765
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012021659119198e-06,
+      "loss": 0.4461,
+      "step": 15766
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010906431845508e-06,
+      "loss": 0.4217,
+      "step": 15767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.009791216994883e-06,
+      "loss": 0.4341,
+      "step": 15768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008676014581338e-06,
+      "loss": 0.4858,
+      "step": 15769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007560824618877e-06,
+      "loss": 0.4414,
+      "step": 15770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00644564712151e-06,
+      "loss": 0.4482,
+      "step": 15771
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005330482103237e-06,
+      "loss": 0.5016,
+      "step": 15772
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004215329578074e-06,
+      "loss": 0.4779,
+      "step": 15773
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00310018956002e-06,
+      "loss": 0.5017,
+      "step": 15774
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001985062063082e-06,
+      "loss": 0.4725,
+      "step": 15775
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.000869947101271e-06,
+      "loss": 0.4377,
+      "step": 15776
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.999754844688586e-06,
+      "loss": 0.4734,
+      "step": 15777
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998639754839036e-06,
+      "loss": 0.4201,
+      "step": 15778
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997524677566627e-06,
+      "loss": 0.421,
+      "step": 15779
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.996409612885367e-06,
+      "loss": 0.4436,
+      "step": 15780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995294560809256e-06,
+      "loss": 0.4633,
+      "step": 15781
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994179521352301e-06,
+      "loss": 0.4146,
+      "step": 15782
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993064494528511e-06,
+      "loss": 0.4476,
+      "step": 15783
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991949480351883e-06,
+      "loss": 0.4032,
+      "step": 15784
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99083447883643e-06,
+      "loss": 0.4607,
+      "step": 15785
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989719489996153e-06,
+      "loss": 0.5222,
+      "step": 15786
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.988604513845051e-06,
+      "loss": 0.4529,
+      "step": 15787
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987489550397133e-06,
+      "loss": 0.4486,
+      "step": 15788
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.986374599666409e-06,
+      "loss": 0.4519,
+      "step": 15789
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985259661666873e-06,
+      "loss": 0.4195,
+      "step": 15790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984144736412531e-06,
+      "loss": 0.4776,
+      "step": 15791
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983029823917392e-06,
+      "loss": 0.4586,
+      "step": 15792
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981914924195451e-06,
+      "loss": 0.4792,
+      "step": 15793
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980800037260716e-06,
+      "loss": 0.4527,
+      "step": 15794
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979685163127194e-06,
+      "loss": 0.4343,
+      "step": 15795
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.978570301808881e-06,
+      "loss": 0.4212,
+      "step": 15796
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97745545331978e-06,
+      "loss": 0.4089,
+      "step": 15797
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976340617673901e-06,
+      "loss": 0.436,
+      "step": 15798
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.975225794885237e-06,
+      "loss": 0.4387,
+      "step": 15799
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974110984967792e-06,
+      "loss": 0.4783,
+      "step": 15800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972996187935577e-06,
+      "loss": 0.4505,
+      "step": 15801
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.971881403802584e-06,
+      "loss": 0.4297,
+      "step": 15802
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970766632582817e-06,
+      "loss": 0.5018,
+      "step": 15803
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969651874290282e-06,
+      "loss": 0.3934,
+      "step": 15804
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.968537128938974e-06,
+      "loss": 0.4643,
+      "step": 15805
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9674223965429e-06,
+      "loss": 0.4763,
+      "step": 15806
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966307677116056e-06,
+      "loss": 0.4052,
+      "step": 15807
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965192970672447e-06,
+      "loss": 0.4894,
+      "step": 15808
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96407827722607e-06,
+      "loss": 0.4453,
+      "step": 15809
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962963596790931e-06,
+      "loss": 0.4625,
+      "step": 15810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961848929381027e-06,
+      "loss": 0.4392,
+      "step": 15811
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.960734275010356e-06,
+      "loss": 0.4173,
+      "step": 15812
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959619633692926e-06,
+      "loss": 0.5069,
+      "step": 15813
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958505005442726e-06,
+      "loss": 0.4268,
+      "step": 15814
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957390390273762e-06,
+      "loss": 0.4557,
+      "step": 15815
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.956275788200035e-06,
+      "loss": 0.4693,
+      "step": 15816
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955161199235543e-06,
+      "loss": 0.454,
+      "step": 15817
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954046623394284e-06,
+      "loss": 0.4145,
+      "step": 15818
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952932060690256e-06,
+      "loss": 0.4641,
+      "step": 15819
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951817511137464e-06,
+      "loss": 0.397,
+      "step": 15820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950702974749899e-06,
+      "loss": 0.4515,
+      "step": 15821
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949588451541565e-06,
+      "loss": 0.4025,
+      "step": 15822
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948473941526463e-06,
+      "loss": 0.4316,
+      "step": 15823
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94735944471858e-06,
+      "loss": 0.4738,
+      "step": 15824
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.946244961131922e-06,
+      "loss": 0.4724,
+      "step": 15825
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945130490780492e-06,
+      "loss": 0.4851,
+      "step": 15826
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944016033678277e-06,
+      "loss": 0.472,
+      "step": 15827
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.942901589839282e-06,
+      "loss": 0.4862,
+      "step": 15828
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941787159277506e-06,
+      "loss": 0.4452,
+      "step": 15829
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.940672742006936e-06,
+      "loss": 0.4474,
+      "step": 15830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.939558338041576e-06,
+      "loss": 0.4618,
+      "step": 15831
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.938443947395428e-06,
+      "loss": 0.4785,
+      "step": 15832
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93732957008248e-06,
+      "loss": 0.4499,
+      "step": 15833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93621520611673e-06,
+      "loss": 0.4298,
+      "step": 15834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935100855512183e-06,
+      "loss": 0.4314,
+      "step": 15835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.933986518282823e-06,
+      "loss": 0.4519,
+      "step": 15836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932872194442651e-06,
+      "loss": 0.471,
+      "step": 15837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93175788400567e-06,
+      "loss": 0.4707,
+      "step": 15838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930643586985867e-06,
+      "loss": 0.4751,
+      "step": 15839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929529303397238e-06,
+      "loss": 0.4685,
+      "step": 15840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.928415033253785e-06,
+      "loss": 0.4361,
+      "step": 15841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.927300776569496e-06,
+      "loss": 0.4612,
+      "step": 15842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92618653335837e-06,
+      "loss": 0.4081,
+      "step": 15843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925072303634403e-06,
+      "loss": 0.4928,
+      "step": 15844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.923958087411587e-06,
+      "loss": 0.4204,
+      "step": 15845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.922843884703917e-06,
+      "loss": 0.4692,
+      "step": 15846
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.921729695525391e-06,
+      "loss": 0.4641,
+      "step": 15847
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.920615519889995e-06,
+      "loss": 0.4541,
+      "step": 15848
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.919501357811732e-06,
+      "loss": 0.4569,
+      "step": 15849
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918387209304596e-06,
+      "loss": 0.389,
+      "step": 15850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.917273074382572e-06,
+      "loss": 0.4627,
+      "step": 15851
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916158953059659e-06,
+      "loss": 0.5069,
+      "step": 15852
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915044845349851e-06,
+      "loss": 0.4673,
+      "step": 15853
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913930751267141e-06,
+      "loss": 0.4119,
+      "step": 15854
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912816670825521e-06,
+      "loss": 0.4421,
+      "step": 15855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911702604038985e-06,
+      "loss": 0.4245,
+      "step": 15856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910588550921528e-06,
+      "loss": 0.4466,
+      "step": 15857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.909474511487136e-06,
+      "loss": 0.486,
+      "step": 15858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.908360485749805e-06,
+      "loss": 0.4033,
+      "step": 15859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907246473723533e-06,
+      "loss": 0.4185,
+      "step": 15860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.906132475422299e-06,
+      "loss": 0.493,
+      "step": 15861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905018490860104e-06,
+      "loss": 0.4053,
+      "step": 15862
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903904520050942e-06,
+      "loss": 0.4437,
+      "step": 15863
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.902790563008797e-06,
+      "loss": 0.4658,
+      "step": 15864
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901676619747666e-06,
+      "loss": 0.4129,
+      "step": 15865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900562690281538e-06,
+      "loss": 0.4291,
+      "step": 15866
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899448774624403e-06,
+      "loss": 0.4839,
+      "step": 15867
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898334872790252e-06,
+      "loss": 0.4705,
+      "step": 15868
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897220984793081e-06,
+      "loss": 0.4261,
+      "step": 15869
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896107110646874e-06,
+      "loss": 0.4509,
+      "step": 15870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89499325036562e-06,
+      "loss": 0.4545,
+      "step": 15871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89387940396332e-06,
+      "loss": 0.4626,
+      "step": 15872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892765571453949e-06,
+      "loss": 0.4927,
+      "step": 15873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891651752851507e-06,
+      "loss": 0.4535,
+      "step": 15874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890537948169984e-06,
+      "loss": 0.4592,
+      "step": 15875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.889424157423366e-06,
+      "loss": 0.4364,
+      "step": 15876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888310380625639e-06,
+      "loss": 0.3947,
+      "step": 15877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887196617790801e-06,
+      "loss": 0.3946,
+      "step": 15878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88608286893283e-06,
+      "loss": 0.4245,
+      "step": 15879
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884969134065724e-06,
+      "loss": 0.4423,
+      "step": 15880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.883855413203468e-06,
+      "loss": 0.4809,
+      "step": 15881
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88274170636005e-06,
+      "loss": 0.4256,
+      "step": 15882
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.881628013549457e-06,
+      "loss": 0.4694,
+      "step": 15883
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880514334785684e-06,
+      "loss": 0.4125,
+      "step": 15884
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879400670082708e-06,
+      "loss": 0.4134,
+      "step": 15885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878287019454526e-06,
+      "loss": 0.4321,
+      "step": 15886
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.877173382915118e-06,
+      "loss": 0.4299,
+      "step": 15887
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87605976047848e-06,
+      "loss": 0.4268,
+      "step": 15888
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87494615215859e-06,
+      "loss": 0.3501,
+      "step": 15889
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873832557969441e-06,
+      "loss": 0.4441,
+      "step": 15890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872718977925023e-06,
+      "loss": 0.448,
+      "step": 15891
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87160541203931e-06,
+      "loss": 0.4689,
+      "step": 15892
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870491860326297e-06,
+      "loss": 0.4183,
+      "step": 15893
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.869378322799975e-06,
+      "loss": 0.4108,
+      "step": 15894
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86826479947432e-06,
+      "loss": 0.4593,
+      "step": 15895
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.867151290363324e-06,
+      "loss": 0.4175,
+      "step": 15896
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866037795480974e-06,
+      "loss": 0.413,
+      "step": 15897
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86492431484125e-06,
+      "loss": 0.4737,
+      "step": 15898
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863810848458138e-06,
+      "loss": 0.5127,
+      "step": 15899
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86269739634563e-06,
+      "loss": 0.4474,
+      "step": 15900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861583958517704e-06,
+      "loss": 0.4255,
+      "step": 15901
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.860470534988346e-06,
+      "loss": 0.4697,
+      "step": 15902
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859357125771544e-06,
+      "loss": 0.4814,
+      "step": 15903
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858243730881278e-06,
+      "loss": 0.4238,
+      "step": 15904
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857130350331535e-06,
+      "loss": 0.4719,
+      "step": 15905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856016984136304e-06,
+      "loss": 0.3905,
+      "step": 15906
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85490363230956e-06,
+      "loss": 0.4738,
+      "step": 15907
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853790294865287e-06,
+      "loss": 0.4537,
+      "step": 15908
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.852676971817479e-06,
+      "loss": 0.4851,
+      "step": 15909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851563663180106e-06,
+      "loss": 0.4422,
+      "step": 15910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850450368967161e-06,
+      "loss": 0.4106,
+      "step": 15911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.849337089192625e-06,
+      "loss": 0.4391,
+      "step": 15912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848223823870478e-06,
+      "loss": 0.4448,
+      "step": 15913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.847110573014702e-06,
+      "loss": 0.4858,
+      "step": 15914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845997336639287e-06,
+      "loss": 0.46,
+      "step": 15915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844884114758206e-06,
+      "loss": 0.4591,
+      "step": 15916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.843770907385445e-06,
+      "loss": 0.4465,
+      "step": 15917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842657714534992e-06,
+      "loss": 0.4631,
+      "step": 15918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841544536220814e-06,
+      "loss": 0.4676,
+      "step": 15919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.840431372456906e-06,
+      "loss": 0.4718,
+      "step": 15920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.839318223257247e-06,
+      "loss": 0.4024,
+      "step": 15921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838205088635811e-06,
+      "loss": 0.4883,
+      "step": 15922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837091968606588e-06,
+      "loss": 0.448,
+      "step": 15923
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835978863183553e-06,
+      "loss": 0.4256,
+      "step": 15924
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834865772380691e-06,
+      "loss": 0.4257,
+      "step": 15925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.833752696211977e-06,
+      "loss": 0.4379,
+      "step": 15926
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832639634691399e-06,
+      "loss": 0.3869,
+      "step": 15927
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831526587832932e-06,
+      "loss": 0.4559,
+      "step": 15928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830413555650552e-06,
+      "loss": 0.45,
+      "step": 15929
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829300538158245e-06,
+      "loss": 0.4616,
+      "step": 15930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82818753536999e-06,
+      "loss": 0.4349,
+      "step": 15931
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827074547299764e-06,
+      "loss": 0.4303,
+      "step": 15932
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825961573961548e-06,
+      "loss": 0.4316,
+      "step": 15933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824848615369323e-06,
+      "loss": 0.451,
+      "step": 15934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82373567153706e-06,
+      "loss": 0.3915,
+      "step": 15935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.822622742478745e-06,
+      "loss": 0.5193,
+      "step": 15936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821509828208356e-06,
+      "loss": 0.5121,
+      "step": 15937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82039692873987e-06,
+      "loss": 0.504,
+      "step": 15938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.81928404408726e-06,
+      "loss": 0.4904,
+      "step": 15939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818171174264515e-06,
+      "loss": 0.4381,
+      "step": 15940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8170583192856e-06,
+      "loss": 0.4587,
+      "step": 15941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8159454791645e-06,
+      "loss": 0.4783,
+      "step": 15942
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.814832653915197e-06,
+      "loss": 0.4461,
+      "step": 15943
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813719843551657e-06,
+      "loss": 0.471,
+      "step": 15944
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.812607048087861e-06,
+      "loss": 0.4932,
+      "step": 15945
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811494267537793e-06,
+      "loss": 0.4481,
+      "step": 15946
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.810381501915417e-06,
+      "loss": 0.4568,
+      "step": 15947
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809268751234718e-06,
+      "loss": 0.424,
+      "step": 15948
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.808156015509672e-06,
+      "loss": 0.4826,
+      "step": 15949
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807043294754252e-06,
+      "loss": 0.4652,
+      "step": 15950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805930588982434e-06,
+      "loss": 0.4753,
+      "step": 15951
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.804817898208198e-06,
+      "loss": 0.4421,
+      "step": 15952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.803705222445511e-06,
+      "loss": 0.4092,
+      "step": 15953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.802592561708358e-06,
+      "loss": 0.4781,
+      "step": 15954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.801479916010709e-06,
+      "loss": 0.4493,
+      "step": 15955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800367285366536e-06,
+      "loss": 0.443,
+      "step": 15956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.799254669789819e-06,
+      "loss": 0.4399,
+      "step": 15957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798142069294532e-06,
+      "loss": 0.4408,
+      "step": 15958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.797029483894644e-06,
+      "loss": 0.4865,
+      "step": 15959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.795916913604137e-06,
+      "loss": 0.4226,
+      "step": 15960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.794804358436978e-06,
+      "loss": 0.4272,
+      "step": 15961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79369181840715e-06,
+      "loss": 0.4693,
+      "step": 15962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792579293528613e-06,
+      "loss": 0.4437,
+      "step": 15963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79146678381535e-06,
+      "loss": 0.4479,
+      "step": 15964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.790354289281335e-06,
+      "loss": 0.4419,
+      "step": 15965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789241809940534e-06,
+      "loss": 0.3742,
+      "step": 15966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788129345806922e-06,
+      "loss": 0.439,
+      "step": 15967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78701689689448e-06,
+      "loss": 0.3871,
+      "step": 15968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785904463217168e-06,
+      "loss": 0.4226,
+      "step": 15969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784792044788965e-06,
+      "loss": 0.471,
+      "step": 15970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.783679641623845e-06,
+      "loss": 0.4855,
+      "step": 15971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782567253735774e-06,
+      "loss": 0.4641,
+      "step": 15972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.781454881138725e-06,
+      "loss": 0.4819,
+      "step": 15973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780342523846675e-06,
+      "loss": 0.4722,
+      "step": 15974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77923018187359e-06,
+      "loss": 0.4713,
+      "step": 15975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77811785523344e-06,
+      "loss": 0.4085,
+      "step": 15976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777005543940203e-06,
+      "loss": 0.4426,
+      "step": 15977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77589324800784e-06,
+      "loss": 0.451,
+      "step": 15978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.774780967450325e-06,
+      "loss": 0.3734,
+      "step": 15979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773668702281636e-06,
+      "loss": 0.4643,
+      "step": 15980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772556452515733e-06,
+      "loss": 0.507,
+      "step": 15981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.771444218166589e-06,
+      "loss": 0.4567,
+      "step": 15982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.770331999248178e-06,
+      "loss": 0.4287,
+      "step": 15983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769219795774461e-06,
+      "loss": 0.4475,
+      "step": 15984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.768107607759415e-06,
+      "loss": 0.4488,
+      "step": 15985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766995435217008e-06,
+      "loss": 0.4176,
+      "step": 15986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765883278161204e-06,
+      "loss": 0.4898,
+      "step": 15987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764771136605976e-06,
+      "loss": 0.4847,
+      "step": 15988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763659010565296e-06,
+      "loss": 0.4639,
+      "step": 15989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.762546900053122e-06,
+      "loss": 0.4619,
+      "step": 15990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76143480508343e-06,
+      "loss": 0.5187,
+      "step": 15991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76032272567019e-06,
+      "loss": 0.4289,
+      "step": 15992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759210661827362e-06,
+      "loss": 0.5153,
+      "step": 15993
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758098613568916e-06,
+      "loss": 0.4414,
+      "step": 15994
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.756986580908827e-06,
+      "loss": 0.4079,
+      "step": 15995
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75587456386105e-06,
+      "loss": 0.4303,
+      "step": 15996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754762562439561e-06,
+      "loss": 0.4377,
+      "step": 15997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.753650576658322e-06,
+      "loss": 0.4294,
+      "step": 15998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752538606531307e-06,
+      "loss": 0.4504,
+      "step": 15999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751426652072472e-06,
+      "loss": 0.3878,
+      "step": 16000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75031471329579e-06,
+      "loss": 0.4441,
+      "step": 16001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749202790215227e-06,
+      "loss": 0.4451,
+      "step": 16002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.748090882844742e-06,
+      "loss": 0.545,
+      "step": 16003
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.746978991198306e-06,
+      "loss": 0.4297,
+      "step": 16004
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74586711528989e-06,
+      "loss": 0.404,
+      "step": 16005
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744755255133447e-06,
+      "loss": 0.4756,
+      "step": 16006
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743643410742952e-06,
+      "loss": 0.4072,
+      "step": 16007
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742531582132368e-06,
+      "loss": 0.452,
+      "step": 16008
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741419769315653e-06,
+      "loss": 0.4422,
+      "step": 16009
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740307972306776e-06,
+      "loss": 0.4426,
+      "step": 16010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739196191119705e-06,
+      "loss": 0.4153,
+      "step": 16011
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7380844257684e-06,
+      "loss": 0.5028,
+      "step": 16012
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736972676266822e-06,
+      "loss": 0.4374,
+      "step": 16013
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.735860942628943e-06,
+      "loss": 0.435,
+      "step": 16014
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734749224868716e-06,
+      "loss": 0.4723,
+      "step": 16015
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733637523000112e-06,
+      "loss": 0.4354,
+      "step": 16016
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732525837037094e-06,
+      "loss": 0.4432,
+      "step": 16017
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73141416699362e-06,
+      "loss": 0.4537,
+      "step": 16018
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730302512883653e-06,
+      "loss": 0.4873,
+      "step": 16019
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.729190874721162e-06,
+      "loss": 0.4183,
+      "step": 16020
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728079252520101e-06,
+      "loss": 0.4245,
+      "step": 16021
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.726967646294439e-06,
+      "loss": 0.4752,
+      "step": 16022
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725856056058135e-06,
+      "loss": 0.4909,
+      "step": 16023
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.724744481825148e-06,
+      "loss": 0.4329,
+      "step": 16024
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723632923609438e-06,
+      "loss": 0.4768,
+      "step": 16025
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722521381424978e-06,
+      "loss": 0.4047,
+      "step": 16026
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721409855285715e-06,
+      "loss": 0.4747,
+      "step": 16027
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.720298345205618e-06,
+      "loss": 0.4116,
+      "step": 16028
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719186851198648e-06,
+      "loss": 0.3932,
+      "step": 16029
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718075373278758e-06,
+      "loss": 0.4959,
+      "step": 16030
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716963911459915e-06,
+      "loss": 0.4326,
+      "step": 16031
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71585246575608e-06,
+      "loss": 0.4325,
+      "step": 16032
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.714741036181207e-06,
+      "loss": 0.4313,
+      "step": 16033
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713629622749258e-06,
+      "loss": 0.4525,
+      "step": 16034
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712518225474191e-06,
+      "loss": 0.4283,
+      "step": 16035
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.711406844369973e-06,
+      "loss": 0.4412,
+      "step": 16036
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710295479450552e-06,
+      "loss": 0.4312,
+      "step": 16037
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.709184130729895e-06,
+      "loss": 0.4427,
+      "step": 16038
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.708072798221961e-06,
+      "loss": 0.4607,
+      "step": 16039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.7069614819407e-06,
+      "loss": 0.4714,
+      "step": 16040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705850181900074e-06,
+      "loss": 0.4717,
+      "step": 16041
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704738898114045e-06,
+      "loss": 0.4222,
+      "step": 16042
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70362763059657e-06,
+      "loss": 0.5043,
+      "step": 16043
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702516379361599e-06,
+      "loss": 0.4006,
+      "step": 16044
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.7014051444231e-06,
+      "loss": 0.4729,
+      "step": 16045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700293925795022e-06,
+      "loss": 0.4434,
+      "step": 16046
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.699182723491324e-06,
+      "loss": 0.5019,
+      "step": 16047
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698071537525968e-06,
+      "loss": 0.4934,
+      "step": 16048
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696960367912905e-06,
+      "loss": 0.4929,
+      "step": 16049
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69584921466609e-06,
+      "loss": 0.4777,
+      "step": 16050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.4282,
+      "step": 16051
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.693626957327042e-06,
+      "loss": 0.4178,
+      "step": 16052
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692515853262717e-06,
+      "loss": 0.4169,
+      "step": 16053
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.691404765620467e-06,
+      "loss": 0.4344,
+      "step": 16054
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.690293694414247e-06,
+      "loss": 0.4869,
+      "step": 16055
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68918263965801e-06,
+      "loss": 0.4771,
+      "step": 16056
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688071601365715e-06,
+      "loss": 0.4523,
+      "step": 16057
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68696057955131e-06,
+      "loss": 0.4639,
+      "step": 16058
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685849574228756e-06,
+      "loss": 0.4192,
+      "step": 16059
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684738585412008e-06,
+      "loss": 0.4388,
+      "step": 16060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68362761311501e-06,
+      "loss": 0.4197,
+      "step": 16061
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.682516657351725e-06,
+      "loss": 0.4128,
+      "step": 16062
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68140571813611e-06,
+      "loss": 0.4256,
+      "step": 16063
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680294795482108e-06,
+      "loss": 0.4769,
+      "step": 16064
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679183889403678e-06,
+      "loss": 0.454,
+      "step": 16065
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.678072999914777e-06,
+      "loss": 0.4707,
+      "step": 16066
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676962127029348e-06,
+      "loss": 0.4939,
+      "step": 16067
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.675851270761346e-06,
+      "loss": 0.4376,
+      "step": 16068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674740431124733e-06,
+      "loss": 0.4709,
+      "step": 16069
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.673629608133451e-06,
+      "loss": 0.4202,
+      "step": 16070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672518801801455e-06,
+      "loss": 0.4316,
+      "step": 16071
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671408012142697e-06,
+      "loss": 0.4927,
+      "step": 16072
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67029723917113e-06,
+      "loss": 0.4851,
+      "step": 16073
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.669186482900705e-06,
+      "loss": 0.4489,
+      "step": 16074
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66807574334537e-06,
+      "loss": 0.4498,
+      "step": 16075
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.666965020519083e-06,
+      "loss": 0.4779,
+      "step": 16076
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665854314435784e-06,
+      "loss": 0.4808,
+      "step": 16077
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66474362510943e-06,
+      "loss": 0.4537,
+      "step": 16078
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663632952553976e-06,
+      "loss": 0.4628,
+      "step": 16079
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.662522296783364e-06,
+      "loss": 0.4751,
+      "step": 16080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661411657811545e-06,
+      "loss": 0.4162,
+      "step": 16081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660301035652476e-06,
+      "loss": 0.4576,
+      "step": 16082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.659190430320094e-06,
+      "loss": 0.4189,
+      "step": 16083
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.658079841828357e-06,
+      "loss": 0.4274,
+      "step": 16084
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.656969270191219e-06,
+      "loss": 0.475,
+      "step": 16085
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655858715422616e-06,
+      "loss": 0.3741,
+      "step": 16086
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.654748177536501e-06,
+      "loss": 0.4231,
+      "step": 16087
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65363765654683e-06,
+      "loss": 0.4711,
+      "step": 16088
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.652527152467541e-06,
+      "loss": 0.4624,
+      "step": 16089
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651416665312588e-06,
+      "loss": 0.422,
+      "step": 16090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650306195095919e-06,
+      "loss": 0.4415,
+      "step": 16091
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649195741831477e-06,
+      "loss": 0.428,
+      "step": 16092
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648085305533212e-06,
+      "loss": 0.4666,
+      "step": 16093
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646974886215076e-06,
+      "loss": 0.4226,
+      "step": 16094
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645864483891006e-06,
+      "loss": 0.4899,
+      "step": 16095
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644754098574955e-06,
+      "loss": 0.4861,
+      "step": 16096
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643643730280872e-06,
+      "loss": 0.4462,
+      "step": 16097
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642533379022694e-06,
+      "loss": 0.4661,
+      "step": 16098
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641423044814375e-06,
+      "loss": 0.4593,
+      "step": 16099
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640312727669861e-06,
+      "loss": 0.4506,
+      "step": 16100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639202427603092e-06,
+      "loss": 0.488,
+      "step": 16101
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.638092144628019e-06,
+      "loss": 0.4095,
+      "step": 16102
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636981878758585e-06,
+      "loss": 0.3945,
+      "step": 16103
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.635871630008732e-06,
+      "loss": 0.5082,
+      "step": 16104
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63476139839241e-06,
+      "loss": 0.4766,
+      "step": 16105
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633651183923561e-06,
+      "loss": 0.4333,
+      "step": 16106
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632540986616132e-06,
+      "loss": 0.469,
+      "step": 16107
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631430806484061e-06,
+      "loss": 0.418,
+      "step": 16108
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630320643541295e-06,
+      "loss": 0.4662,
+      "step": 16109
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.629210497801783e-06,
+      "loss": 0.4933,
+      "step": 16110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.628100369279461e-06,
+      "loss": 0.4159,
+      "step": 16111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.626990257988275e-06,
+      "loss": 0.4916,
+      "step": 16112
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625880163942172e-06,
+      "loss": 0.4578,
+      "step": 16113
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624770087155086e-06,
+      "loss": 0.4581,
+      "step": 16114
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.623660027640966e-06,
+      "loss": 0.4304,
+      "step": 16115
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622549985413757e-06,
+      "loss": 0.4765,
+      "step": 16116
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621439960487395e-06,
+      "loss": 0.4719,
+      "step": 16117
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620329952875822e-06,
+      "loss": 0.4354,
+      "step": 16118
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619219962592986e-06,
+      "loss": 0.4595,
+      "step": 16119
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61810998965282e-06,
+      "loss": 0.438,
+      "step": 16120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61700003406927e-06,
+      "loss": 0.4446,
+      "step": 16121
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.615890095856282e-06,
+      "loss": 0.4196,
+      "step": 16122
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61478017502779e-06,
+      "loss": 0.4747,
+      "step": 16123
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613670271597733e-06,
+      "loss": 0.4058,
+      "step": 16124
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.612560385580062e-06,
+      "loss": 0.4749,
+      "step": 16125
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.611450516988703e-06,
+      "loss": 0.4569,
+      "step": 16126
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.610340665837606e-06,
+      "loss": 0.4344,
+      "step": 16127
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60923083214071e-06,
+      "loss": 0.4755,
+      "step": 16128
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.608121015911952e-06,
+      "loss": 0.4542,
+      "step": 16129
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607011217165269e-06,
+      "loss": 0.4325,
+      "step": 16130
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.605901435914608e-06,
+      "loss": 0.446,
+      "step": 16131
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.604791672173898e-06,
+      "loss": 0.4959,
+      "step": 16132
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.603681925957085e-06,
+      "loss": 0.4261,
+      "step": 16133
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.602572197278107e-06,
+      "loss": 0.4795,
+      "step": 16134
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601462486150897e-06,
+      "loss": 0.4596,
+      "step": 16135
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.600352792589397e-06,
+      "loss": 0.4638,
+      "step": 16136
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599243116607548e-06,
+      "loss": 0.4594,
+      "step": 16137
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.598133458219278e-06,
+      "loss": 0.3738,
+      "step": 16138
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.597023817438535e-06,
+      "loss": 0.4934,
+      "step": 16139
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.595914194279253e-06,
+      "loss": 0.4873,
+      "step": 16140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594804588755362e-06,
+      "loss": 0.4162,
+      "step": 16141
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.593695000880806e-06,
+      "loss": 0.4901,
+      "step": 16142
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59258543066952e-06,
+      "loss": 0.4974,
+      "step": 16143
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591475878135442e-06,
+      "loss": 0.4462,
+      "step": 16144
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590366343292503e-06,
+      "loss": 0.4619,
+      "step": 16145
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58925682615464e-06,
+      "loss": 0.4821,
+      "step": 16146
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588147326735794e-06,
+      "loss": 0.4692,
+      "step": 16147
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587037845049893e-06,
+      "loss": 0.4553,
+      "step": 16148
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.585928381110879e-06,
+      "loss": 0.4619,
+      "step": 16149
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584818934932686e-06,
+      "loss": 0.4673,
+      "step": 16150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583709506529241e-06,
+      "loss": 0.4624,
+      "step": 16151
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582600095914485e-06,
+      "loss": 0.424,
+      "step": 16152
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581490703102356e-06,
+      "loss": 0.4281,
+      "step": 16153
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580381328106779e-06,
+      "loss": 0.4509,
+      "step": 16154
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57927197094169e-06,
+      "loss": 0.4568,
+      "step": 16155
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57816263162103e-06,
+      "loss": 0.4547,
+      "step": 16156
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577053310158723e-06,
+      "loss": 0.421,
+      "step": 16157
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575944006568706e-06,
+      "loss": 0.4545,
+      "step": 16158
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574834720864916e-06,
+      "loss": 0.399,
+      "step": 16159
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573725453061278e-06,
+      "loss": 0.4064,
+      "step": 16160
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.572616203171729e-06,
+      "loss": 0.3947,
+      "step": 16161
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571506971210202e-06,
+      "loss": 0.3806,
+      "step": 16162
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570397757190624e-06,
+      "loss": 0.3774,
+      "step": 16163
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569288561126934e-06,
+      "loss": 0.407,
+      "step": 16164
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.568179383033059e-06,
+      "loss": 0.4786,
+      "step": 16165
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56707022292293e-06,
+      "loss": 0.4324,
+      "step": 16166
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565961080810478e-06,
+      "loss": 0.4219,
+      "step": 16167
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564851956709639e-06,
+      "loss": 0.473,
+      "step": 16168
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563742850634335e-06,
+      "loss": 0.4531,
+      "step": 16169
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562633762598504e-06,
+      "loss": 0.4031,
+      "step": 16170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561524692616076e-06,
+      "loss": 0.4461,
+      "step": 16171
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.560415640700974e-06,
+      "loss": 0.477,
+      "step": 16172
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559306606867132e-06,
+      "loss": 0.4807,
+      "step": 16173
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558197591128484e-06,
+      "loss": 0.4531,
+      "step": 16174
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.557088593498953e-06,
+      "loss": 0.4843,
+      "step": 16175
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555979613992467e-06,
+      "loss": 0.4014,
+      "step": 16176
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554870652622965e-06,
+      "loss": 0.4658,
+      "step": 16177
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553761709404362e-06,
+      "loss": 0.4138,
+      "step": 16178
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.552652784350592e-06,
+      "loss": 0.4234,
+      "step": 16179
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551543877475588e-06,
+      "loss": 0.465,
+      "step": 16180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550434988793274e-06,
+      "loss": 0.428,
+      "step": 16181
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549326118317576e-06,
+      "loss": 0.4359,
+      "step": 16182
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.548217266062422e-06,
+      "loss": 0.5098,
+      "step": 16183
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547108432041745e-06,
+      "loss": 0.4425,
+      "step": 16184
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.545999616269466e-06,
+      "loss": 0.4265,
+      "step": 16185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54489081875951e-06,
+      "loss": 0.4721,
+      "step": 16186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.543782039525812e-06,
+      "loss": 0.4616,
+      "step": 16187
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542673278582289e-06,
+      "loss": 0.5156,
+      "step": 16188
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54156453594287e-06,
+      "loss": 0.4623,
+      "step": 16189
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540455811621486e-06,
+      "loss": 0.4506,
+      "step": 16190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53934710563206e-06,
+      "loss": 0.4731,
+      "step": 16191
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538238417988511e-06,
+      "loss": 0.4512,
+      "step": 16192
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.537129748704773e-06,
+      "loss": 0.4841,
+      "step": 16193
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536021097794765e-06,
+      "loss": 0.4826,
+      "step": 16194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534912465272414e-06,
+      "loss": 0.4595,
+      "step": 16195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.533803851151648e-06,
+      "loss": 0.4491,
+      "step": 16196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532695255446384e-06,
+      "loss": 0.3999,
+      "step": 16197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53158667817055e-06,
+      "loss": 0.4674,
+      "step": 16198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.530478119338073e-06,
+      "loss": 0.4848,
+      "step": 16199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529369578962868e-06,
+      "loss": 0.4567,
+      "step": 16200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528261057058868e-06,
+      "loss": 0.4522,
+      "step": 16201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.527152553639991e-06,
+      "loss": 0.4422,
+      "step": 16202
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526044068720159e-06,
+      "loss": 0.4778,
+      "step": 16203
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.524935602313293e-06,
+      "loss": 0.5043,
+      "step": 16204
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523827154433324e-06,
+      "loss": 0.4733,
+      "step": 16205
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522718725094165e-06,
+      "loss": 0.4576,
+      "step": 16206
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521610314309743e-06,
+      "loss": 0.4787,
+      "step": 16207
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52050192209398e-06,
+      "loss": 0.4041,
+      "step": 16208
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519393548460792e-06,
+      "loss": 0.4499,
+      "step": 16209
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518285193424104e-06,
+      "loss": 0.4693,
+      "step": 16210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517176856997842e-06,
+      "loss": 0.4366,
+      "step": 16211
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.516068539195918e-06,
+      "loss": 0.4194,
+      "step": 16212
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514960240032255e-06,
+      "loss": 0.4522,
+      "step": 16213
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.513851959520779e-06,
+      "loss": 0.4494,
+      "step": 16214
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5127436976754e-06,
+      "loss": 0.4038,
+      "step": 16215
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511635454510045e-06,
+      "loss": 0.4734,
+      "step": 16216
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510527230038634e-06,
+      "loss": 0.4797,
+      "step": 16217
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509419024275086e-06,
+      "loss": 0.4293,
+      "step": 16218
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.508310837233315e-06,
+      "loss": 0.464,
+      "step": 16219
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.507202668927244e-06,
+      "loss": 0.4869,
+      "step": 16220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506094519370793e-06,
+      "loss": 0.3761,
+      "step": 16221
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504986388577878e-06,
+      "loss": 0.4212,
+      "step": 16222
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503878276562415e-06,
+      "loss": 0.445,
+      "step": 16223
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502770183338329e-06,
+      "loss": 0.481,
+      "step": 16224
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50166210891953e-06,
+      "loss": 0.4613,
+      "step": 16225
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500554053319937e-06,
+      "loss": 0.4406,
+      "step": 16226
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.499446016553475e-06,
+      "loss": 0.449,
+      "step": 16227
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498337998634051e-06,
+      "loss": 0.4486,
+      "step": 16228
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497229999575587e-06,
+      "loss": 0.4601,
+      "step": 16229
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496122019392e-06,
+      "loss": 0.458,
+      "step": 16230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.495014058097201e-06,
+      "loss": 0.4578,
+      "step": 16231
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.493906115705112e-06,
+      "loss": 0.4752,
+      "step": 16232
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.492798192229647e-06,
+      "loss": 0.4695,
+      "step": 16233
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.49169028768472e-06,
+      "loss": 0.4595,
+      "step": 16234
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490582402084245e-06,
+      "loss": 0.4734,
+      "step": 16235
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.489474535442146e-06,
+      "loss": 0.4044,
+      "step": 16236
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488366687772325e-06,
+      "loss": 0.4393,
+      "step": 16237
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.487258859088706e-06,
+      "loss": 0.4869,
+      "step": 16238
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486151049405204e-06,
+      "loss": 0.4283,
+      "step": 16239
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.485043258735723e-06,
+      "loss": 0.4566,
+      "step": 16240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.483935487094185e-06,
+      "loss": 0.4237,
+      "step": 16241
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482827734494505e-06,
+      "loss": 0.4501,
+      "step": 16242
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.481720000950591e-06,
+      "loss": 0.4718,
+      "step": 16243
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480612286476359e-06,
+      "loss": 0.4678,
+      "step": 16244
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479504591085727e-06,
+      "loss": 0.5374,
+      "step": 16245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.478396914792595e-06,
+      "loss": 0.466,
+      "step": 16246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477289257610886e-06,
+      "loss": 0.4501,
+      "step": 16247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47618161955451e-06,
+      "loss": 0.3943,
+      "step": 16248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.475074000637379e-06,
+      "loss": 0.5096,
+      "step": 16249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.473966400873402e-06,
+      "loss": 0.4431,
+      "step": 16250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472858820276494e-06,
+      "loss": 0.4121,
+      "step": 16251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471751258860562e-06,
+      "loss": 0.466,
+      "step": 16252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.470643716639519e-06,
+      "loss": 0.4456,
+      "step": 16253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46953619362728e-06,
+      "loss": 0.4463,
+      "step": 16254
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.468428689837754e-06,
+      "loss": 0.4314,
+      "step": 16255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467321205284845e-06,
+      "loss": 0.4961,
+      "step": 16256
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466213739982468e-06,
+      "loss": 0.4619,
+      "step": 16257
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465106293944535e-06,
+      "loss": 0.4572,
+      "step": 16258
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463998867184952e-06,
+      "loss": 0.4373,
+      "step": 16259
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462891459717627e-06,
+      "loss": 0.4859,
+      "step": 16260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461784071556476e-06,
+      "loss": 0.3845,
+      "step": 16261
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.460676702715399e-06,
+      "loss": 0.4676,
+      "step": 16262
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.459569353208309e-06,
+      "loss": 0.425,
+      "step": 16263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.458462023049116e-06,
+      "loss": 0.4999,
+      "step": 16264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.457354712251727e-06,
+      "loss": 0.4998,
+      "step": 16265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456247420830045e-06,
+      "loss": 0.4865,
+      "step": 16266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455140148797987e-06,
+      "loss": 0.4416,
+      "step": 16267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.454032896169452e-06,
+      "loss": 0.4951,
+      "step": 16268
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452925662958351e-06,
+      "loss": 0.5173,
+      "step": 16269
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451818449178592e-06,
+      "loss": 0.5159,
+      "step": 16270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450711254844079e-06,
+      "loss": 0.4215,
+      "step": 16271
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.449604079968716e-06,
+      "loss": 0.4365,
+      "step": 16272
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448496924566418e-06,
+      "loss": 0.4673,
+      "step": 16273
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44738978865108e-06,
+      "loss": 0.4461,
+      "step": 16274
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44628267223662e-06,
+      "loss": 0.4722,
+      "step": 16275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445175575336935e-06,
+      "loss": 0.4459,
+      "step": 16276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444068497965926e-06,
+      "loss": 0.4349,
+      "step": 16277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442961440137507e-06,
+      "loss": 0.4714,
+      "step": 16278
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.441854401865582e-06,
+      "loss": 0.4538,
+      "step": 16279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44074738316405e-06,
+      "loss": 0.429,
+      "step": 16280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43964038404682e-06,
+      "loss": 0.4184,
+      "step": 16281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438533404527795e-06,
+      "loss": 0.4694,
+      "step": 16282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437426444620875e-06,
+      "loss": 0.4497,
+      "step": 16283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436319504339966e-06,
+      "loss": 0.4437,
+      "step": 16284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.435212583698976e-06,
+      "loss": 0.4693,
+      "step": 16285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434105682711801e-06,
+      "loss": 0.4814,
+      "step": 16286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432998801392345e-06,
+      "loss": 0.461,
+      "step": 16287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431891939754515e-06,
+      "loss": 0.4665,
+      "step": 16288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.430785097812207e-06,
+      "loss": 0.409,
+      "step": 16289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.429678275579325e-06,
+      "loss": 0.4257,
+      "step": 16290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428571473069775e-06,
+      "loss": 0.4577,
+      "step": 16291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.427464690297457e-06,
+      "loss": 0.4575,
+      "step": 16292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426357927276265e-06,
+      "loss": 0.4518,
+      "step": 16293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425251184020106e-06,
+      "loss": 0.4813,
+      "step": 16294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424144460542887e-06,
+      "loss": 0.4919,
+      "step": 16295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423037756858497e-06,
+      "loss": 0.4552,
+      "step": 16296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42193107298084e-06,
+      "loss": 0.4731,
+      "step": 16297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420824408923822e-06,
+      "loss": 0.4807,
+      "step": 16298
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.419717764701333e-06,
+      "loss": 0.4352,
+      "step": 16299
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418611140327277e-06,
+      "loss": 0.4406,
+      "step": 16300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417504535815558e-06,
+      "loss": 0.4542,
+      "step": 16301
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41639795118007e-06,
+      "loss": 0.4775,
+      "step": 16302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.415291386434707e-06,
+      "loss": 0.4533,
+      "step": 16303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41418484159338e-06,
+      "loss": 0.4455,
+      "step": 16304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.413078316669976e-06,
+      "loss": 0.3907,
+      "step": 16305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411971811678398e-06,
+      "loss": 0.5028,
+      "step": 16306
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.410865326632546e-06,
+      "loss": 0.4951,
+      "step": 16307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40975886154631e-06,
+      "loss": 0.4199,
+      "step": 16308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408652416433591e-06,
+      "loss": 0.4442,
+      "step": 16309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407545991308293e-06,
+      "loss": 0.4405,
+      "step": 16310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406439586184299e-06,
+      "loss": 0.4393,
+      "step": 16311
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40533320107552e-06,
+      "loss": 0.4708,
+      "step": 16312
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404226835995843e-06,
+      "loss": 0.4215,
+      "step": 16313
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.403120490959164e-06,
+      "loss": 0.448,
+      "step": 16314
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402014165979379e-06,
+      "loss": 0.4411,
+      "step": 16315
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.400907861070393e-06,
+      "loss": 0.4121,
+      "step": 16316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.399801576246089e-06,
+      "loss": 0.4397,
+      "step": 16317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.398695311520366e-06,
+      "loss": 0.4776,
+      "step": 16318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397589066907123e-06,
+      "loss": 0.4395,
+      "step": 16319
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396482842420249e-06,
+      "loss": 0.4884,
+      "step": 16320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395376638073638e-06,
+      "loss": 0.4346,
+      "step": 16321
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394270453881192e-06,
+      "loss": 0.4784,
+      "step": 16322
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393164289856797e-06,
+      "loss": 0.4205,
+      "step": 16323
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392058146014345e-06,
+      "loss": 0.4414,
+      "step": 16324
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390952022367733e-06,
+      "loss": 0.5104,
+      "step": 16325
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.38984591893086e-06,
+      "loss": 0.445,
+      "step": 16326
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.388739835717607e-06,
+      "loss": 0.4675,
+      "step": 16327
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387633772741872e-06,
+      "loss": 0.4603,
+      "step": 16328
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38652773001755e-06,
+      "loss": 0.4358,
+      "step": 16329
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.385421707558527e-06,
+      "loss": 0.454,
+      "step": 16330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384315705378699e-06,
+      "loss": 0.4497,
+      "step": 16331
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.383209723491958e-06,
+      "loss": 0.4734,
+      "step": 16332
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.382103761912193e-06,
+      "loss": 0.4215,
+      "step": 16333
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380997820653292e-06,
+      "loss": 0.3897,
+      "step": 16334
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379891899729153e-06,
+      "loss": 0.4808,
+      "step": 16335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378785999153658e-06,
+      "loss": 0.4021,
+      "step": 16336
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377680118940707e-06,
+      "loss": 0.4316,
+      "step": 16337
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.376574259104181e-06,
+      "loss": 0.4159,
+      "step": 16338
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375468419657975e-06,
+      "loss": 0.4534,
+      "step": 16339
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374362600615974e-06,
+      "loss": 0.4789,
+      "step": 16340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373256801992073e-06,
+      "loss": 0.4617,
+      "step": 16341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.372151023800156e-06,
+      "loss": 0.4409,
+      "step": 16342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.371045266054114e-06,
+      "loss": 0.4797,
+      "step": 16343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.369939528767837e-06,
+      "loss": 0.4529,
+      "step": 16344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.368833811955206e-06,
+      "loss": 0.4295,
+      "step": 16345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367728115630114e-06,
+      "loss": 0.5016,
+      "step": 16346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366622439806453e-06,
+      "loss": 0.4196,
+      "step": 16347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3655167844981e-06,
+      "loss": 0.493,
+      "step": 16348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.364411149718951e-06,
+      "loss": 0.4457,
+      "step": 16349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363305535482891e-06,
+      "loss": 0.454,
+      "step": 16350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.362199941803801e-06,
+      "loss": 0.4238,
+      "step": 16351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361094368695572e-06,
+      "loss": 0.4017,
+      "step": 16352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359988816172095e-06,
+      "loss": 0.4596,
+      "step": 16353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.358883284247246e-06,
+      "loss": 0.48,
+      "step": 16354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.357777772934914e-06,
+      "loss": 0.4167,
+      "step": 16355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356672282248989e-06,
+      "loss": 0.4368,
+      "step": 16356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355566812203348e-06,
+      "loss": 0.4925,
+      "step": 16357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354461362811878e-06,
+      "loss": 0.4621,
+      "step": 16358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.353355934088473e-06,
+      "loss": 0.497,
+      "step": 16359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352250526047006e-06,
+      "loss": 0.4331,
+      "step": 16360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.351145138701362e-06,
+      "loss": 0.4766,
+      "step": 16361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.350039772065429e-06,
+      "loss": 0.512,
+      "step": 16362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348934426153091e-06,
+      "loss": 0.4374,
+      "step": 16363
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347829100978229e-06,
+      "loss": 0.3836,
+      "step": 16364
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346723796554722e-06,
+      "loss": 0.4306,
+      "step": 16365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345618512896464e-06,
+      "loss": 0.4261,
+      "step": 16366
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.344513250017323e-06,
+      "loss": 0.4232,
+      "step": 16367
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343408007931189e-06,
+      "loss": 0.4173,
+      "step": 16368
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342302786651946e-06,
+      "loss": 0.4602,
+      "step": 16369
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341197586193472e-06,
+      "loss": 0.4665,
+      "step": 16370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340092406569648e-06,
+      "loss": 0.4486,
+      "step": 16371
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33898724779436e-06,
+      "loss": 0.438,
+      "step": 16372
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33788210988148e-06,
+      "loss": 0.4525,
+      "step": 16373
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.336776992844898e-06,
+      "loss": 0.4505,
+      "step": 16374
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33567189669849e-06,
+      "loss": 0.434,
+      "step": 16375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.334566821456134e-06,
+      "loss": 0.4597,
+      "step": 16376
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33346176713171e-06,
+      "loss": 0.4742,
+      "step": 16377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332356733739107e-06,
+      "loss": 0.455,
+      "step": 16378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331251721292189e-06,
+      "loss": 0.4644,
+      "step": 16379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330146729804847e-06,
+      "loss": 0.39,
+      "step": 16380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.329041759290957e-06,
+      "loss": 0.4211,
+      "step": 16381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.327936809764393e-06,
+      "loss": 0.5188,
+      "step": 16382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326831881239034e-06,
+      "loss": 0.4804,
+      "step": 16383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325726973728764e-06,
+      "loss": 0.4257,
+      "step": 16384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.324622087247457e-06,
+      "loss": 0.4193,
+      "step": 16385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323517221808988e-06,
+      "loss": 0.4418,
+      "step": 16386
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322412377427242e-06,
+      "loss": 0.4594,
+      "step": 16387
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321307554116085e-06,
+      "loss": 0.4702,
+      "step": 16388
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3202027518894e-06,
+      "loss": 0.4332,
+      "step": 16389
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.319097970761066e-06,
+      "loss": 0.4638,
+      "step": 16390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317993210744953e-06,
+      "loss": 0.4907,
+      "step": 16391
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316888471854937e-06,
+      "loss": 0.5065,
+      "step": 16392
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315783754104901e-06,
+      "loss": 0.4317,
+      "step": 16393
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314679057508712e-06,
+      "loss": 0.4335,
+      "step": 16394
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.313574382080245e-06,
+      "loss": 0.4325,
+      "step": 16395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312469727833386e-06,
+      "loss": 0.4422,
+      "step": 16396
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311365094782e-06,
+      "loss": 0.4532,
+      "step": 16397
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.310260482939958e-06,
+      "loss": 0.4703,
+      "step": 16398
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309155892321141e-06,
+      "loss": 0.5255,
+      "step": 16399
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308051322939424e-06,
+      "loss": 0.4327,
+      "step": 16400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306946774808676e-06,
+      "loss": 0.4586,
+      "step": 16401
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305842247942769e-06,
+      "loss": 0.3826,
+      "step": 16402
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304737742355582e-06,
+      "loss": 0.4822,
+      "step": 16403
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.30363325806098e-06,
+      "loss": 0.4133,
+      "step": 16404
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302528795072838e-06,
+      "loss": 0.4662,
+      "step": 16405
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301424353405035e-06,
+      "loss": 0.4072,
+      "step": 16406
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300319933071435e-06,
+      "loss": 0.4418,
+      "step": 16407
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.29921553408591e-06,
+      "loss": 0.4353,
+      "step": 16408
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298111156462336e-06,
+      "loss": 0.4547,
+      "step": 16409
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297006800214578e-06,
+      "loss": 0.4962,
+      "step": 16410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295902465356512e-06,
+      "loss": 0.4185,
+      "step": 16411
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294798151902007e-06,
+      "loss": 0.4888,
+      "step": 16412
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293693859864931e-06,
+      "loss": 0.4552,
+      "step": 16413
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292589589259155e-06,
+      "loss": 0.4811,
+      "step": 16414
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291485340098554e-06,
+      "loss": 0.438,
+      "step": 16415
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290381112396989e-06,
+      "loss": 0.4366,
+      "step": 16416
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289276906168334e-06,
+      "loss": 0.4509,
+      "step": 16417
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.288172721426458e-06,
+      "loss": 0.5026,
+      "step": 16418
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287068558185225e-06,
+      "loss": 0.4813,
+      "step": 16419
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285964416458507e-06,
+      "loss": 0.4484,
+      "step": 16420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.284860296260178e-06,
+      "loss": 0.4972,
+      "step": 16421
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.283756197604092e-06,
+      "loss": 0.439,
+      "step": 16422
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.282652120504128e-06,
+      "loss": 0.436,
+      "step": 16423
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.281548064974152e-06,
+      "loss": 0.4091,
+      "step": 16424
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.280444031028025e-06,
+      "loss": 0.4435,
+      "step": 16425
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.279340018679616e-06,
+      "loss": 0.4753,
+      "step": 16426
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278236027942797e-06,
+      "loss": 0.4935,
+      "step": 16427
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.277132058831429e-06,
+      "loss": 0.4361,
+      "step": 16428
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276028111359376e-06,
+      "loss": 0.3983,
+      "step": 16429
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27492418554051e-06,
+      "loss": 0.4777,
+      "step": 16430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27382028138869e-06,
+      "loss": 0.4673,
+      "step": 16431
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272716398917783e-06,
+      "loss": 0.4047,
+      "step": 16432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27161253814166e-06,
+      "loss": 0.4426,
+      "step": 16433
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270508699074177e-06,
+      "loss": 0.4301,
+      "step": 16434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269404881729199e-06,
+      "loss": 0.4715,
+      "step": 16435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268301086120594e-06,
+      "loss": 0.4403,
+      "step": 16436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.267197312262228e-06,
+      "loss": 0.4463,
+      "step": 16437
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26609356016796e-06,
+      "loss": 0.4268,
+      "step": 16438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26498982985165e-06,
+      "loss": 0.4374,
+      "step": 16439
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263886121327169e-06,
+      "loss": 0.452,
+      "step": 16440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.262782434608372e-06,
+      "loss": 0.4751,
+      "step": 16441
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261678769709128e-06,
+      "loss": 0.4366,
+      "step": 16442
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260575126643296e-06,
+      "loss": 0.462,
+      "step": 16443
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259471505424736e-06,
+      "loss": 0.48,
+      "step": 16444
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25836790606731e-06,
+      "loss": 0.4277,
+      "step": 16445
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.257264328584885e-06,
+      "loss": 0.5041,
+      "step": 16446
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256160772991313e-06,
+      "loss": 0.4553,
+      "step": 16447
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.255057239300462e-06,
+      "loss": 0.4223,
+      "step": 16448
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253953727526193e-06,
+      "loss": 0.4645,
+      "step": 16449
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252850237682358e-06,
+      "loss": 0.3964,
+      "step": 16450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.251746769782821e-06,
+      "loss": 0.4032,
+      "step": 16451
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250643323841448e-06,
+      "loss": 0.4352,
+      "step": 16452
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249539899872087e-06,
+      "loss": 0.4934,
+      "step": 16453
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248436497888606e-06,
+      "loss": 0.4373,
+      "step": 16454
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247333117904862e-06,
+      "loss": 0.4323,
+      "step": 16455
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24622975993471e-06,
+      "loss": 0.4502,
+      "step": 16456
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245126423992008e-06,
+      "loss": 0.4361,
+      "step": 16457
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244023110090621e-06,
+      "loss": 0.4225,
+      "step": 16458
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.242919818244402e-06,
+      "loss": 0.4544,
+      "step": 16459
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241816548467205e-06,
+      "loss": 0.4877,
+      "step": 16460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240713300772893e-06,
+      "loss": 0.4391,
+      "step": 16461
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23961007517532e-06,
+      "loss": 0.3937,
+      "step": 16462
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23850687168834e-06,
+      "loss": 0.4979,
+      "step": 16463
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237403690325818e-06,
+      "loss": 0.4749,
+      "step": 16464
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.236300531101602e-06,
+      "loss": 0.4997,
+      "step": 16465
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.235197394029548e-06,
+      "loss": 0.4846,
+      "step": 16466
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.234094279123517e-06,
+      "loss": 0.4381,
+      "step": 16467
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.232991186397358e-06,
+      "loss": 0.4154,
+      "step": 16468
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23188811586493e-06,
+      "loss": 0.4148,
+      "step": 16469
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.230785067540086e-06,
+      "loss": 0.446,
+      "step": 16470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22968204143668e-06,
+      "loss": 0.4587,
+      "step": 16471
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228579037568568e-06,
+      "loss": 0.4894,
+      "step": 16472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227476055949599e-06,
+      "loss": 0.5106,
+      "step": 16473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.226373096593636e-06,
+      "loss": 0.4488,
+      "step": 16474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225270159514522e-06,
+      "loss": 0.4755,
+      "step": 16475
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224167244726114e-06,
+      "loss": 0.4456,
+      "step": 16476
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.223064352242269e-06,
+      "loss": 0.4636,
+      "step": 16477
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22196148207683e-06,
+      "loss": 0.4708,
+      "step": 16478
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.220858634243658e-06,
+      "loss": 0.4411,
+      "step": 16479
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2197558087566e-06,
+      "loss": 0.5235,
+      "step": 16480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21865300562951e-06,
+      "loss": 0.4857,
+      "step": 16481
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.217550224876233e-06,
+      "loss": 0.4224,
+      "step": 16482
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.216447466510633e-06,
+      "loss": 0.4432,
+      "step": 16483
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215344730546547e-06,
+      "loss": 0.4341,
+      "step": 16484
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21424201699783e-06,
+      "loss": 0.458,
+      "step": 16485
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21313932587834e-06,
+      "loss": 0.4434,
+      "step": 16486
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.212036657201914e-06,
+      "loss": 0.4526,
+      "step": 16487
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210934010982408e-06,
+      "loss": 0.4645,
+      "step": 16488
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209831387233675e-06,
+      "loss": 0.4537,
+      "step": 16489
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208728785969556e-06,
+      "loss": 0.4759,
+      "step": 16490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207626207203907e-06,
+      "loss": 0.4417,
+      "step": 16491
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206523650950573e-06,
+      "loss": 0.4403,
+      "step": 16492
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2054211172234e-06,
+      "loss": 0.4624,
+      "step": 16493
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204318606036237e-06,
+      "loss": 0.4702,
+      "step": 16494
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203216117402939e-06,
+      "loss": 0.455,
+      "step": 16495
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.202113651337343e-06,
+      "loss": 0.4743,
+      "step": 16496
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2010112078533e-06,
+      "loss": 0.4,
+      "step": 16497
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19990878696466e-06,
+      "loss": 0.441,
+      "step": 16498
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198806388685262e-06,
+      "loss": 0.4729,
+      "step": 16499
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.197704013028958e-06,
+      "loss": 0.4515,
+      "step": 16500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196601660009595e-06,
+      "loss": 0.4737,
+      "step": 16501
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195499329641015e-06,
+      "loss": 0.4252,
+      "step": 16502
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.194397021937061e-06,
+      "loss": 0.4795,
+      "step": 16503
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193294736911588e-06,
+      "loss": 0.4871,
+      "step": 16504
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19219247457843e-06,
+      "loss": 0.4457,
+      "step": 16505
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.191090234951435e-06,
+      "loss": 0.4151,
+      "step": 16506
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18998801804445e-06,
+      "loss": 0.4389,
+      "step": 16507
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188885823871317e-06,
+      "loss": 0.4405,
+      "step": 16508
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187783652445877e-06,
+      "loss": 0.42,
+      "step": 16509
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.186681503781977e-06,
+      "loss": 0.4725,
+      "step": 16510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185579377893462e-06,
+      "loss": 0.4546,
+      "step": 16511
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18447727479417e-06,
+      "loss": 0.4413,
+      "step": 16512
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.183375194497942e-06,
+      "loss": 0.4964,
+      "step": 16513
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182273137018629e-06,
+      "loss": 0.4625,
+      "step": 16514
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181171102370062e-06,
+      "loss": 0.4778,
+      "step": 16515
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18006909056609e-06,
+      "loss": 0.4392,
+      "step": 16516
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178967101620552e-06,
+      "loss": 0.431,
+      "step": 16517
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.177865135547289e-06,
+      "loss": 0.4005,
+      "step": 16518
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.17676319236014e-06,
+      "loss": 0.4471,
+      "step": 16519
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.175661272072953e-06,
+      "loss": 0.4566,
+      "step": 16520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174559374699555e-06,
+      "loss": 0.4519,
+      "step": 16521
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1734575002538e-06,
+      "loss": 0.4798,
+      "step": 16522
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.172355648749519e-06,
+      "loss": 0.3947,
+      "step": 16523
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171253820200552e-06,
+      "loss": 0.4373,
+      "step": 16524
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.170152014620738e-06,
+      "loss": 0.467,
+      "step": 16525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169050232023923e-06,
+      "loss": 0.4213,
+      "step": 16526
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167948472423934e-06,
+      "loss": 0.4068,
+      "step": 16527
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.166846735834616e-06,
+      "loss": 0.4377,
+      "step": 16528
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.165745022269811e-06,
+      "loss": 0.4734,
+      "step": 16529
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.164643331743345e-06,
+      "loss": 0.4552,
+      "step": 16530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.163541664269061e-06,
+      "loss": 0.4391,
+      "step": 16531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162440019860803e-06,
+      "loss": 0.4528,
+      "step": 16532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161338398532396e-06,
+      "loss": 0.4577,
+      "step": 16533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.160236800297681e-06,
+      "loss": 0.4555,
+      "step": 16534
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1591352251705e-06,
+      "loss": 0.4302,
+      "step": 16535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.158033673164679e-06,
+      "loss": 0.4907,
+      "step": 16536
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156932144294057e-06,
+      "loss": 0.4536,
+      "step": 16537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155830638572477e-06,
+      "loss": 0.47,
+      "step": 16538
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154729156013763e-06,
+      "loss": 0.4452,
+      "step": 16539
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.153627696631753e-06,
+      "loss": 0.4731,
+      "step": 16540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152526260440286e-06,
+      "loss": 0.4301,
+      "step": 16541
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15142484745319e-06,
+      "loss": 0.3863,
+      "step": 16542
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.150323457684302e-06,
+      "loss": 0.4363,
+      "step": 16543
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149222091147454e-06,
+      "loss": 0.4726,
+      "step": 16544
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148120747856486e-06,
+      "loss": 0.4842,
+      "step": 16545
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147019427825217e-06,
+      "loss": 0.4498,
+      "step": 16546
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14591813106749e-06,
+      "loss": 0.4298,
+      "step": 16547
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144816857597138e-06,
+      "loss": 0.4402,
+      "step": 16548
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14371560742799e-06,
+      "loss": 0.4441,
+      "step": 16549
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142614380573874e-06,
+      "loss": 0.4468,
+      "step": 16550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.141513177048633e-06,
+      "loss": 0.4724,
+      "step": 16551
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140411996866082e-06,
+      "loss": 0.4378,
+      "step": 16552
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139310840040064e-06,
+      "loss": 0.4372,
+      "step": 16553
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138209706584408e-06,
+      "loss": 0.4658,
+      "step": 16554
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13710859651294e-06,
+      "loss": 0.4985,
+      "step": 16555
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.136007509839492e-06,
+      "loss": 0.4031,
+      "step": 16556
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134906446577896e-06,
+      "loss": 0.3809,
+      "step": 16557
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133805406741977e-06,
+      "loss": 0.5408,
+      "step": 16558
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13270439034557e-06,
+      "loss": 0.4198,
+      "step": 16559
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1316033974025e-06,
+      "loss": 0.4394,
+      "step": 16560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.130502427926593e-06,
+      "loss": 0.4432,
+      "step": 16561
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12940148193168e-06,
+      "loss": 0.4411,
+      "step": 16562
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128300559431594e-06,
+      "loss": 0.4415,
+      "step": 16563
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.127199660440152e-06,
+      "loss": 0.3693,
+      "step": 16564
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12609878497119e-06,
+      "loss": 0.4362,
+      "step": 16565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124997933038534e-06,
+      "loss": 0.3995,
+      "step": 16566
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123897104656005e-06,
+      "loss": 0.4604,
+      "step": 16567
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122796299837434e-06,
+      "loss": 0.4533,
+      "step": 16568
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12169551859665e-06,
+      "loss": 0.4568,
+      "step": 16569
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120594760947476e-06,
+      "loss": 0.4304,
+      "step": 16570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119494026903731e-06,
+      "loss": 0.4388,
+      "step": 16571
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118393316479254e-06,
+      "loss": 0.4567,
+      "step": 16572
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.117292629687856e-06,
+      "loss": 0.4499,
+      "step": 16573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11619196654337e-06,
+      "loss": 0.4488,
+      "step": 16574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11509132705962e-06,
+      "loss": 0.4809,
+      "step": 16575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113990711250427e-06,
+      "loss": 0.4988,
+      "step": 16576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112890119129615e-06,
+      "loss": 0.4512,
+      "step": 16577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111789550711013e-06,
+      "loss": 0.4985,
+      "step": 16578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.110689006008435e-06,
+      "loss": 0.4412,
+      "step": 16579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10958848503571e-06,
+      "loss": 0.422,
+      "step": 16580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10848798780666e-06,
+      "loss": 0.4298,
+      "step": 16581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10738751433511e-06,
+      "loss": 0.4311,
+      "step": 16582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106287064634875e-06,
+      "loss": 0.4109,
+      "step": 16583
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.105186638719781e-06,
+      "loss": 0.4906,
+      "step": 16584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10408623660365e-06,
+      "loss": 0.438,
+      "step": 16585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.102985858300302e-06,
+      "loss": 0.448,
+      "step": 16586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101885503823557e-06,
+      "loss": 0.4375,
+      "step": 16587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10078517318724e-06,
+      "loss": 0.4403,
+      "step": 16588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099684866405163e-06,
+      "loss": 0.4372,
+      "step": 16589
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098584583491152e-06,
+      "loss": 0.4982,
+      "step": 16590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.097484324459028e-06,
+      "loss": 0.4396,
+      "step": 16591
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096384089322601e-06,
+      "loss": 0.4897,
+      "step": 16592
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0952838780957e-06,
+      "loss": 0.4392,
+      "step": 16593
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094183690792143e-06,
+      "loss": 0.4448,
+      "step": 16594
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09308352742574e-06,
+      "loss": 0.412,
+      "step": 16595
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091983388010318e-06,
+      "loss": 0.4735,
+      "step": 16596
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090883272559694e-06,
+      "loss": 0.4533,
+      "step": 16597
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089783181087679e-06,
+      "loss": 0.4213,
+      "step": 16598
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088683113608094e-06,
+      "loss": 0.3972,
+      "step": 16599
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087583070134759e-06,
+      "loss": 0.4372,
+      "step": 16600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.086483050681486e-06,
+      "loss": 0.4363,
+      "step": 16601
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085383055262092e-06,
+      "loss": 0.4288,
+      "step": 16602
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084283083890399e-06,
+      "loss": 0.4568,
+      "step": 16603
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083183136580211e-06,
+      "loss": 0.4516,
+      "step": 16604
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082083213345352e-06,
+      "loss": 0.4804,
+      "step": 16605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08098331419964e-06,
+      "loss": 0.4616,
+      "step": 16606
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.079883439156882e-06,
+      "loss": 0.4608,
+      "step": 16607
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078783588230892e-06,
+      "loss": 0.4437,
+      "step": 16608
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.077683761435493e-06,
+      "loss": 0.4534,
+      "step": 16609
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07658395878449e-06,
+      "loss": 0.4721,
+      "step": 16610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075484180291702e-06,
+      "loss": 0.418,
+      "step": 16611
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.074384425970941e-06,
+      "loss": 0.4376,
+      "step": 16612
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073284695836018e-06,
+      "loss": 0.4268,
+      "step": 16613
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.072184989900746e-06,
+      "loss": 0.4746,
+      "step": 16614
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.071085308178944e-06,
+      "loss": 0.4591,
+      "step": 16615
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.069985650684413e-06,
+      "loss": 0.4496,
+      "step": 16616
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06888601743097e-06,
+      "loss": 0.4354,
+      "step": 16617
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.067786408432429e-06,
+      "loss": 0.4443,
+      "step": 16618
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0666868237026e-06,
+      "loss": 0.4473,
+      "step": 16619
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065587263255291e-06,
+      "loss": 0.4152,
+      "step": 16620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.064487727104315e-06,
+      "loss": 0.4269,
+      "step": 16621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063388215263483e-06,
+      "loss": 0.5002,
+      "step": 16622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.062288727746601e-06,
+      "loss": 0.4085,
+      "step": 16623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06118926456748e-06,
+      "loss": 0.5159,
+      "step": 16624
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060089825739936e-06,
+      "loss": 0.4571,
+      "step": 16625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.058990411277767e-06,
+      "loss": 0.4356,
+      "step": 16626
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057891021194789e-06,
+      "loss": 0.4088,
+      "step": 16627
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056791655504811e-06,
+      "loss": 0.4779,
+      "step": 16628
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055692314221635e-06,
+      "loss": 0.4985,
+      "step": 16629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054592997359071e-06,
+      "loss": 0.4062,
+      "step": 16630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053493704930933e-06,
+      "loss": 0.487,
+      "step": 16631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.052394436951019e-06,
+      "loss": 0.4514,
+      "step": 16632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.051295193433141e-06,
+      "loss": 0.4684,
+      "step": 16633
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050195974391108e-06,
+      "loss": 0.4635,
+      "step": 16634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.4544,
+      "step": 16635
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047997609789783e-06,
+      "loss": 0.4421,
+      "step": 16636
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04689846425811e-06,
+      "loss": 0.4858,
+      "step": 16637
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0457993432575e-06,
+      "loss": 0.4471,
+      "step": 16638
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.044700246801758e-06,
+      "loss": 0.4331,
+      "step": 16639
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043601174904694e-06,
+      "loss": 0.4595,
+      "step": 16640
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.042502127580104e-06,
+      "loss": 0.4523,
+      "step": 16641
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041403104841797e-06,
+      "loss": 0.4294,
+      "step": 16642
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.040304106703582e-06,
+      "loss": 0.4296,
+      "step": 16643
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039205133179256e-06,
+      "loss": 0.4338,
+      "step": 16644
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038106184282619e-06,
+      "loss": 0.501,
+      "step": 16645
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037007260027484e-06,
+      "loss": 0.4531,
+      "step": 16646
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.035908360427643e-06,
+      "loss": 0.4293,
+      "step": 16647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034809485496904e-06,
+      "loss": 0.4411,
+      "step": 16648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.033710635249069e-06,
+      "loss": 0.4758,
+      "step": 16649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032611809697937e-06,
+      "loss": 0.474,
+      "step": 16650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031513008857312e-06,
+      "loss": 0.4691,
+      "step": 16651
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030414232740995e-06,
+      "loss": 0.4192,
+      "step": 16652
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029315481362783e-06,
+      "loss": 0.4816,
+      "step": 16653
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02821675473648e-06,
+      "loss": 0.4716,
+      "step": 16654
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.027118052875883e-06,
+      "loss": 0.4596,
+      "step": 16655
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0260193757948e-06,
+      "loss": 0.4306,
+      "step": 16656
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024920723507019e-06,
+      "loss": 0.4437,
+      "step": 16657
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023822096026344e-06,
+      "loss": 0.4559,
+      "step": 16658
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022723493366576e-06,
+      "loss": 0.4325,
+      "step": 16659
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021624915541513e-06,
+      "loss": 0.3788,
+      "step": 16660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.020526362564948e-06,
+      "loss": 0.4398,
+      "step": 16661
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019427834450687e-06,
+      "loss": 0.4248,
+      "step": 16662
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01832933121252e-06,
+      "loss": 0.4445,
+      "step": 16663
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017230852864248e-06,
+      "loss": 0.463,
+      "step": 16664
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.016132399419672e-06,
+      "loss": 0.3874,
+      "step": 16665
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.015033970892578e-06,
+      "loss": 0.4239,
+      "step": 16666
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01393556729677e-06,
+      "loss": 0.3759,
+      "step": 16667
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012837188646045e-06,
+      "loss": 0.4149,
+      "step": 16668
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011738834954194e-06,
+      "loss": 0.4366,
+      "step": 16669
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.010640506235016e-06,
+      "loss": 0.4834,
+      "step": 16670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009542202502306e-06,
+      "loss": 0.4036,
+      "step": 16671
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008443923769853e-06,
+      "loss": 0.4439,
+      "step": 16672
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.007345670051456e-06,
+      "loss": 0.4346,
+      "step": 16673
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006247441360915e-06,
+      "loss": 0.4813,
+      "step": 16674
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005149237712014e-06,
+      "loss": 0.4982,
+      "step": 16675
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.004051059118549e-06,
+      "loss": 0.4006,
+      "step": 16676
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002952905594318e-06,
+      "loss": 0.4422,
+      "step": 16677
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001854777153106e-06,
+      "loss": 0.4963,
+      "step": 16678
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00075667380871e-06,
+      "loss": 0.448,
+      "step": 16679
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.999658595574928e-06,
+      "loss": 0.4225,
+      "step": 16680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.998560542465541e-06,
+      "loss": 0.4216,
+      "step": 16681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997462514494346e-06,
+      "loss": 0.4345,
+      "step": 16682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.996364511675138e-06,
+      "loss": 0.4424,
+      "step": 16683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995266534021698e-06,
+      "loss": 0.4515,
+      "step": 16684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.994168581547827e-06,
+      "loss": 0.4201,
+      "step": 16685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99307065426731e-06,
+      "loss": 0.4973,
+      "step": 16686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991972752193937e-06,
+      "loss": 0.4773,
+      "step": 16687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.990874875341497e-06,
+      "loss": 0.4131,
+      "step": 16688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.989777023723787e-06,
+      "loss": 0.4487,
+      "step": 16689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988679197354583e-06,
+      "loss": 0.4632,
+      "step": 16690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987581396247685e-06,
+      "loss": 0.5096,
+      "step": 16691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.986483620416876e-06,
+      "loss": 0.3967,
+      "step": 16692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985385869875947e-06,
+      "loss": 0.4431,
+      "step": 16693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.984288144638683e-06,
+      "loss": 0.4101,
+      "step": 16694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983190444718873e-06,
+      "loss": 0.4469,
+      "step": 16695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.982092770130305e-06,
+      "loss": 0.4456,
+      "step": 16696
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.980995120886765e-06,
+      "loss": 0.5066,
+      "step": 16697
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979897497002036e-06,
+      "loss": 0.4309,
+      "step": 16698
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978799898489914e-06,
+      "loss": 0.4884,
+      "step": 16699
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977702325364171e-06,
+      "loss": 0.4503,
+      "step": 16700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.976604777638604e-06,
+      "loss": 0.3943,
+      "step": 16701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975507255326997e-06,
+      "loss": 0.4667,
+      "step": 16702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.974409758443127e-06,
+      "loss": 0.4907,
+      "step": 16703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.973312287000785e-06,
+      "loss": 0.4404,
+      "step": 16704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972214841013756e-06,
+      "loss": 0.4645,
+      "step": 16705
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971117420495822e-06,
+      "loss": 0.4007,
+      "step": 16706
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970020025460765e-06,
+      "loss": 0.4244,
+      "step": 16707
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.968922655922375e-06,
+      "loss": 0.4863,
+      "step": 16708
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967825311894423e-06,
+      "loss": 0.4366,
+      "step": 16709
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966727993390701e-06,
+      "loss": 0.4046,
+      "step": 16710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.965630700424994e-06,
+      "loss": 0.4661,
+      "step": 16711
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964533433011076e-06,
+      "loss": 0.429,
+      "step": 16712
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96343619116273e-06,
+      "loss": 0.4703,
+      "step": 16713
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.962338974893745e-06,
+      "loss": 0.4496,
+      "step": 16714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.961241784217892e-06,
+      "loss": 0.4893,
+      "step": 16715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960144619148957e-06,
+      "loss": 0.4597,
+      "step": 16716
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959047479700721e-06,
+      "loss": 0.4956,
+      "step": 16717
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.957950365886963e-06,
+      "loss": 0.4263,
+      "step": 16718
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956853277721461e-06,
+      "loss": 0.4819,
+      "step": 16719
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955756215218e-06,
+      "loss": 0.4109,
+      "step": 16720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954659178390351e-06,
+      "loss": 0.4264,
+      "step": 16721
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.953562167252299e-06,
+      "loss": 0.4715,
+      "step": 16722
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952465181817624e-06,
+      "loss": 0.4869,
+      "step": 16723
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951368222100095e-06,
+      "loss": 0.4075,
+      "step": 16724
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950271288113496e-06,
+      "loss": 0.4505,
+      "step": 16725
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.949174379871611e-06,
+      "loss": 0.4219,
+      "step": 16726
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948077497388204e-06,
+      "loss": 0.4563,
+      "step": 16727
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946980640677062e-06,
+      "loss": 0.478,
+      "step": 16728
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.945883809751956e-06,
+      "loss": 0.4499,
+      "step": 16729
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944787004626668e-06,
+      "loss": 0.4351,
+      "step": 16730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943690225314968e-06,
+      "loss": 0.4263,
+      "step": 16731
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942593471830635e-06,
+      "loss": 0.4486,
+      "step": 16732
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941496744187444e-06,
+      "loss": 0.4532,
+      "step": 16733
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940400042399165e-06,
+      "loss": 0.4876,
+      "step": 16734
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939303366479577e-06,
+      "loss": 0.4237,
+      "step": 16735
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93820671644246e-06,
+      "loss": 0.45,
+      "step": 16736
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937110092301578e-06,
+      "loss": 0.4996,
+      "step": 16737
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.936013494070709e-06,
+      "loss": 0.4071,
+      "step": 16738
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93491692176363e-06,
+      "loss": 0.426,
+      "step": 16739
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933820375394104e-06,
+      "loss": 0.414,
+      "step": 16740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.932723854975911e-06,
+      "loss": 0.4149,
+      "step": 16741
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.931627360522826e-06,
+      "loss": 0.4737,
+      "step": 16742
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930530892048615e-06,
+      "loss": 0.4382,
+      "step": 16743
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92943444956705e-06,
+      "loss": 0.4466,
+      "step": 16744
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92833803309191e-06,
+      "loss": 0.4368,
+      "step": 16745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927241642636953e-06,
+      "loss": 0.4415,
+      "step": 16746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926145278215959e-06,
+      "loss": 0.4178,
+      "step": 16747
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.9250489398427e-06,
+      "loss": 0.425,
+      "step": 16748
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923952627530941e-06,
+      "loss": 0.4771,
+      "step": 16749
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92285634129445e-06,
+      "loss": 0.4422,
+      "step": 16750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921760081147005e-06,
+      "loss": 0.4434,
+      "step": 16751
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.920663847102366e-06,
+      "loss": 0.4346,
+      "step": 16752
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919567639174307e-06,
+      "loss": 0.4495,
+      "step": 16753
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.918471457376595e-06,
+      "loss": 0.461,
+      "step": 16754
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917375301722998e-06,
+      "loss": 0.4448,
+      "step": 16755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.916279172227282e-06,
+      "loss": 0.4982,
+      "step": 16756
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91518306890322e-06,
+      "loss": 0.4592,
+      "step": 16757
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914086991764572e-06,
+      "loss": 0.4035,
+      "step": 16758
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91299094082511e-06,
+      "loss": 0.4523,
+      "step": 16759
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.9118949160986e-06,
+      "loss": 0.479,
+      "step": 16760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910798917598803e-06,
+      "loss": 0.4193,
+      "step": 16761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90970294533949e-06,
+      "loss": 0.4215,
+      "step": 16762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.908606999334425e-06,
+      "loss": 0.4389,
+      "step": 16763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907511079597374e-06,
+      "loss": 0.3711,
+      "step": 16764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.9064151861421e-06,
+      "loss": 0.4491,
+      "step": 16765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.905319318982368e-06,
+      "loss": 0.4074,
+      "step": 16766
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.904223478131948e-06,
+      "loss": 0.4102,
+      "step": 16767
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.903127663604592e-06,
+      "loss": 0.4376,
+      "step": 16768
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902031875414073e-06,
+      "loss": 0.4486,
+      "step": 16769
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900936113574153e-06,
+      "loss": 0.4826,
+      "step": 16770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899840378098589e-06,
+      "loss": 0.4652,
+      "step": 16771
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.898744669001146e-06,
+      "loss": 0.4712,
+      "step": 16772
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.897648986295594e-06,
+      "loss": 0.4665,
+      "step": 16773
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896553329995684e-06,
+      "loss": 0.429,
+      "step": 16774
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895457700115184e-06,
+      "loss": 0.4599,
+      "step": 16775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.894362096667855e-06,
+      "loss": 0.42,
+      "step": 16776
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89326651966745e-06,
+      "loss": 0.5135,
+      "step": 16777
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892170969127739e-06,
+      "loss": 0.4372,
+      "step": 16778
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89107544506248e-06,
+      "loss": 0.4599,
+      "step": 16779
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88997994748543e-06,
+      "loss": 0.4354,
+      "step": 16780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888884476410348e-06,
+      "loss": 0.404,
+      "step": 16781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887789031851001e-06,
+      "loss": 0.455,
+      "step": 16782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886693613821136e-06,
+      "loss": 0.4385,
+      "step": 16783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885598222334518e-06,
+      "loss": 0.4477,
+      "step": 16784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.884502857404909e-06,
+      "loss": 0.4612,
+      "step": 16785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.883407519046061e-06,
+      "loss": 0.5285,
+      "step": 16786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88231220727173e-06,
+      "loss": 0.5173,
+      "step": 16787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.881216922095683e-06,
+      "loss": 0.4568,
+      "step": 16788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.880121663531664e-06,
+      "loss": 0.5128,
+      "step": 16789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.879026431593437e-06,
+      "loss": 0.4455,
+      "step": 16790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.877931226294756e-06,
+      "loss": 0.423,
+      "step": 16791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876836047649382e-06,
+      "loss": 0.4832,
+      "step": 16792
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87574089567106e-06,
+      "loss": 0.428,
+      "step": 16793
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874645770373557e-06,
+      "loss": 0.4624,
+      "step": 16794
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873550671770618e-06,
+      "loss": 0.4625,
+      "step": 16795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872455599876004e-06,
+      "loss": 0.4571,
+      "step": 16796
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871360554703467e-06,
+      "loss": 0.3965,
+      "step": 16797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.870265536266758e-06,
+      "loss": 0.4354,
+      "step": 16798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.869170544579632e-06,
+      "loss": 0.4154,
+      "step": 16799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868075579655847e-06,
+      "loss": 0.4272,
+      "step": 16800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866980641509152e-06,
+      "loss": 0.4607,
+      "step": 16801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.8658857301533e-06,
+      "loss": 0.4476,
+      "step": 16802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86479084560204e-06,
+      "loss": 0.408,
+      "step": 16803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86369598786913e-06,
+      "loss": 0.4207,
+      "step": 16804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862601156968313e-06,
+      "loss": 0.4813,
+      "step": 16805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861506352913349e-06,
+      "loss": 0.5181,
+      "step": 16806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.860411575717988e-06,
+      "loss": 0.4497,
+      "step": 16807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.859316825395972e-06,
+      "loss": 0.4632,
+      "step": 16808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858222101961056e-06,
+      "loss": 0.4395,
+      "step": 16809
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857127405426996e-06,
+      "loss": 0.4036,
+      "step": 16810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85603273580753e-06,
+      "loss": 0.4594,
+      "step": 16811
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.854938093116416e-06,
+      "loss": 0.4664,
+      "step": 16812
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853843477367402e-06,
+      "loss": 0.4463,
+      "step": 16813
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852748888574228e-06,
+      "loss": 0.4487,
+      "step": 16814
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.851654326750651e-06,
+      "loss": 0.4506,
+      "step": 16815
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.850559791910418e-06,
+      "loss": 0.4247,
+      "step": 16816
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849465284067272e-06,
+      "loss": 0.4466,
+      "step": 16817
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848370803234963e-06,
+      "loss": 0.4123,
+      "step": 16818
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84727634942724e-06,
+      "loss": 0.5447,
+      "step": 16819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846181922657842e-06,
+      "loss": 0.4787,
+      "step": 16820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84508752294052e-06,
+      "loss": 0.4967,
+      "step": 16821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843993150289023e-06,
+      "loss": 0.4687,
+      "step": 16822
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.842898804717092e-06,
+      "loss": 0.4022,
+      "step": 16823
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84180448623847e-06,
+      "loss": 0.4645,
+      "step": 16824
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.840710194866909e-06,
+      "loss": 0.4059,
+      "step": 16825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839615930616144e-06,
+      "loss": 0.4542,
+      "step": 16826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.838521693499927e-06,
+      "loss": 0.4112,
+      "step": 16827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837427483531998e-06,
+      "loss": 0.4075,
+      "step": 16828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8363333007261e-06,
+      "loss": 0.4743,
+      "step": 16829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835239145095976e-06,
+      "loss": 0.4022,
+      "step": 16830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834145016655373e-06,
+      "loss": 0.4417,
+      "step": 16831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833050915418026e-06,
+      "loss": 0.4532,
+      "step": 16832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831956841397685e-06,
+      "loss": 0.4457,
+      "step": 16833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830862794608087e-06,
+      "loss": 0.45,
+      "step": 16834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82976877506297e-06,
+      "loss": 0.4215,
+      "step": 16835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82867478277608e-06,
+      "loss": 0.4378,
+      "step": 16836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827580817761157e-06,
+      "loss": 0.4358,
+      "step": 16837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.826486880031944e-06,
+      "loss": 0.4346,
+      "step": 16838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825392969602172e-06,
+      "loss": 0.4351,
+      "step": 16839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824299086485585e-06,
+      "loss": 0.4884,
+      "step": 16840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82320523069593e-06,
+      "loss": 0.4614,
+      "step": 16841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822111402246935e-06,
+      "loss": 0.5046,
+      "step": 16842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.821017601152344e-06,
+      "loss": 0.4765,
+      "step": 16843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.819923827425895e-06,
+      "loss": 0.4416,
+      "step": 16844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818830081081323e-06,
+      "loss": 0.4318,
+      "step": 16845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.817736362132367e-06,
+      "loss": 0.4538,
+      "step": 16846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816642670592766e-06,
+      "loss": 0.4457,
+      "step": 16847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815549006476257e-06,
+      "loss": 0.498,
+      "step": 16848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81445536979657e-06,
+      "loss": 0.4532,
+      "step": 16849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81336176056745e-06,
+      "loss": 0.4707,
+      "step": 16850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812268178802627e-06,
+      "loss": 0.4736,
+      "step": 16851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811174624515836e-06,
+      "loss": 0.4628,
+      "step": 16852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810081097720819e-06,
+      "loss": 0.4724,
+      "step": 16853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808987598431303e-06,
+      "loss": 0.4856,
+      "step": 16854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807894126661025e-06,
+      "loss": 0.4109,
+      "step": 16855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806800682423725e-06,
+      "loss": 0.4293,
+      "step": 16856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805707265733124e-06,
+      "loss": 0.4398,
+      "step": 16857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804613876602966e-06,
+      "loss": 0.4629,
+      "step": 16858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803520515046981e-06,
+      "loss": 0.4828,
+      "step": 16859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8024271810789e-06,
+      "loss": 0.4847,
+      "step": 16860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801333874712457e-06,
+      "loss": 0.4851,
+      "step": 16861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800240595961385e-06,
+      "loss": 0.4568,
+      "step": 16862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.799147344839413e-06,
+      "loss": 0.4574,
+      "step": 16863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798054121360275e-06,
+      "loss": 0.4473,
+      "step": 16864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796960925537703e-06,
+      "loss": 0.4833,
+      "step": 16865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795867757385419e-06,
+      "loss": 0.4228,
+      "step": 16866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79477461691716e-06,
+      "loss": 0.4299,
+      "step": 16867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.793681504146659e-06,
+      "loss": 0.4157,
+      "step": 16868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79258841908764e-06,
+      "loss": 0.4045,
+      "step": 16869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791495361753834e-06,
+      "loss": 0.385,
+      "step": 16870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790402332158972e-06,
+      "loss": 0.3977,
+      "step": 16871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789309330316777e-06,
+      "loss": 0.4888,
+      "step": 16872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788216356240982e-06,
+      "loss": 0.4171,
+      "step": 16873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.787123409945313e-06,
+      "loss": 0.3932,
+      "step": 16874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786030491443503e-06,
+      "loss": 0.4592,
+      "step": 16875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784937600749267e-06,
+      "loss": 0.3745,
+      "step": 16876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78384473787634e-06,
+      "loss": 0.4267,
+      "step": 16877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782751902838451e-06,
+      "loss": 0.4502,
+      "step": 16878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.781659095649318e-06,
+      "loss": 0.4731,
+      "step": 16879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780566316322674e-06,
+      "loss": 0.4356,
+      "step": 16880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.779473564872242e-06,
+      "loss": 0.4507,
+      "step": 16881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778380841311745e-06,
+      "loss": 0.4295,
+      "step": 16882
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.777288145654906e-06,
+      "loss": 0.4585,
+      "step": 16883
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776195477915459e-06,
+      "loss": 0.4122,
+      "step": 16884
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.775102838107117e-06,
+      "loss": 0.4699,
+      "step": 16885
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774010226243608e-06,
+      "loss": 0.4074,
+      "step": 16886
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772917642338658e-06,
+      "loss": 0.4806,
+      "step": 16887
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771825086405982e-06,
+      "loss": 0.4816,
+      "step": 16888
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77073255845931e-06,
+      "loss": 0.4949,
+      "step": 16889
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.769640058512364e-06,
+      "loss": 0.4104,
+      "step": 16890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768547586578863e-06,
+      "loss": 0.4539,
+      "step": 16891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767455142672527e-06,
+      "loss": 0.4791,
+      "step": 16892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.766362726807085e-06,
+      "loss": 0.4707,
+      "step": 16893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.765270338996244e-06,
+      "loss": 0.461,
+      "step": 16894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764177979253738e-06,
+      "loss": 0.3808,
+      "step": 16895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.763085647593281e-06,
+      "loss": 0.4471,
+      "step": 16896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761993344028595e-06,
+      "loss": 0.4793,
+      "step": 16897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.760901068573394e-06,
+      "loss": 0.4404,
+      "step": 16898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759808821241406e-06,
+      "loss": 0.4542,
+      "step": 16899
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75871660204634e-06,
+      "loss": 0.4516,
+      "step": 16900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757624411001922e-06,
+      "loss": 0.4895,
+      "step": 16901
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756532248121868e-06,
+      "loss": 0.4736,
+      "step": 16902
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755440113419892e-06,
+      "loss": 0.4575,
+      "step": 16903
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754348006909711e-06,
+      "loss": 0.4587,
+      "step": 16904
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753255928605052e-06,
+      "loss": 0.4525,
+      "step": 16905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.752163878519619e-06,
+      "loss": 0.4354,
+      "step": 16906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751071856667136e-06,
+      "loss": 0.4621,
+      "step": 16907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749979863061317e-06,
+      "loss": 0.4221,
+      "step": 16908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748887897715873e-06,
+      "loss": 0.5298,
+      "step": 16909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747795960644525e-06,
+      "loss": 0.4474,
+      "step": 16910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.746704051860987e-06,
+      "loss": 0.4259,
+      "step": 16911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745612171378974e-06,
+      "loss": 0.4117,
+      "step": 16912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744520319212194e-06,
+      "loss": 0.5135,
+      "step": 16913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.743428495374366e-06,
+      "loss": 0.4118,
+      "step": 16914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742336699879206e-06,
+      "loss": 0.4871,
+      "step": 16915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74124493274042e-06,
+      "loss": 0.4613,
+      "step": 16916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740153193971726e-06,
+      "loss": 0.4225,
+      "step": 16917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.739061483586836e-06,
+      "loss": 0.4391,
+      "step": 16918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737969801599457e-06,
+      "loss": 0.4334,
+      "step": 16919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.736878148023304e-06,
+      "loss": 0.4317,
+      "step": 16920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735786522872091e-06,
+      "loss": 0.4438,
+      "step": 16921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734694926159525e-06,
+      "loss": 0.4604,
+      "step": 16922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733603357899315e-06,
+      "loss": 0.4401,
+      "step": 16923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732511818105179e-06,
+      "loss": 0.3925,
+      "step": 16924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.731420306790817e-06,
+      "loss": 0.4997,
+      "step": 16925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730328823969942e-06,
+      "loss": 0.481,
+      "step": 16926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72923736965627e-06,
+      "loss": 0.4503,
+      "step": 16927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.7281459438635e-06,
+      "loss": 0.4912,
+      "step": 16928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.727054546605343e-06,
+      "loss": 0.4627,
+      "step": 16929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725963177895512e-06,
+      "loss": 0.4078,
+      "step": 16930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724871837747708e-06,
+      "loss": 0.4268,
+      "step": 16931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723780526175641e-06,
+      "loss": 0.4575,
+      "step": 16932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72268924319302e-06,
+      "loss": 0.4602,
+      "step": 16933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721597988813548e-06,
+      "loss": 0.4412,
+      "step": 16934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720506763050933e-06,
+      "loss": 0.429,
+      "step": 16935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719415565918882e-06,
+      "loss": 0.4106,
+      "step": 16936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718324397431097e-06,
+      "loss": 0.4549,
+      "step": 16937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717233257601288e-06,
+      "loss": 0.4586,
+      "step": 16938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716142146443157e-06,
+      "loss": 0.4745,
+      "step": 16939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.715051063970406e-06,
+      "loss": 0.4322,
+      "step": 16940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.713960010196742e-06,
+      "loss": 0.4341,
+      "step": 16941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71286898513587e-06,
+      "loss": 0.4361,
+      "step": 16942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71177798880149e-06,
+      "loss": 0.5055,
+      "step": 16943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710687021207308e-06,
+      "loss": 0.4371,
+      "step": 16944
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709596082367028e-06,
+      "loss": 0.4139,
+      "step": 16945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708505172294344e-06,
+      "loss": 0.4486,
+      "step": 16946
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.707414291002963e-06,
+      "loss": 0.5108,
+      "step": 16947
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70632343850659e-06,
+      "loss": 0.4706,
+      "step": 16948
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705232614818925e-06,
+      "loss": 0.4494,
+      "step": 16949
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704141819953663e-06,
+      "loss": 0.4934,
+      "step": 16950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703051053924508e-06,
+      "loss": 0.4279,
+      "step": 16951
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.701960316745164e-06,
+      "loss": 0.4617,
+      "step": 16952
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700869608429324e-06,
+      "loss": 0.438,
+      "step": 16953
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.699778928990692e-06,
+      "loss": 0.5138,
+      "step": 16954
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.698688278442969e-06,
+      "loss": 0.4222,
+      "step": 16955
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697597656799844e-06,
+      "loss": 0.4323,
+      "step": 16956
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696507064075021e-06,
+      "loss": 0.4546,
+      "step": 16957
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695416500282205e-06,
+      "loss": 0.4142,
+      "step": 16958
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.694325965435084e-06,
+      "loss": 0.4495,
+      "step": 16959
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693235459547353e-06,
+      "loss": 0.482,
+      "step": 16960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692144982632721e-06,
+      "loss": 0.4658,
+      "step": 16961
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.691054534704872e-06,
+      "loss": 0.4096,
+      "step": 16962
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.689964115777507e-06,
+      "loss": 0.4375,
+      "step": 16963
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688873725864328e-06,
+      "loss": 0.4014,
+      "step": 16964
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68778336497902e-06,
+      "loss": 0.4558,
+      "step": 16965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.686693033135281e-06,
+      "loss": 0.4288,
+      "step": 16966
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685602730346814e-06,
+      "loss": 0.4766,
+      "step": 16967
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684512456627298e-06,
+      "loss": 0.4306,
+      "step": 16968
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68342221199044e-06,
+      "loss": 0.3939,
+      "step": 16969
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68233199644993e-06,
+      "loss": 0.4239,
+      "step": 16970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681241810019456e-06,
+      "loss": 0.4252,
+      "step": 16971
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680151652712716e-06,
+      "loss": 0.3958,
+      "step": 16972
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.679061524543403e-06,
+      "loss": 0.4357,
+      "step": 16973
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677971425525203e-06,
+      "loss": 0.426,
+      "step": 16974
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676881355671815e-06,
+      "loss": 0.4492,
+      "step": 16975
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67579131499693e-06,
+      "loss": 0.4561,
+      "step": 16976
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674701303514232e-06,
+      "loss": 0.4294,
+      "step": 16977
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673611321237415e-06,
+      "loss": 0.5003,
+      "step": 16978
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.672521368180174e-06,
+      "loss": 0.4257,
+      "step": 16979
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671431444356194e-06,
+      "loss": 0.4441,
+      "step": 16980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.670341549779163e-06,
+      "loss": 0.4045,
+      "step": 16981
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669251684462773e-06,
+      "loss": 0.4761,
+      "step": 16982
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.668161848420715e-06,
+      "loss": 0.4547,
+      "step": 16983
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667072041666672e-06,
+      "loss": 0.3884,
+      "step": 16984
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.665982264214337e-06,
+      "loss": 0.4377,
+      "step": 16985
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.664892516077396e-06,
+      "loss": 0.5131,
+      "step": 16986
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663802797269534e-06,
+      "loss": 0.4811,
+      "step": 16987
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662713107804439e-06,
+      "loss": 0.498,
+      "step": 16988
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.6616234476958e-06,
+      "loss": 0.4808,
+      "step": 16989
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660533816957304e-06,
+      "loss": 0.4798,
+      "step": 16990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659444215602627e-06,
+      "loss": 0.4232,
+      "step": 16991
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65835464364547e-06,
+      "loss": 0.4527,
+      "step": 16992
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657265101099503e-06,
+      "loss": 0.4645,
+      "step": 16993
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.656175587978418e-06,
+      "loss": 0.3641,
+      "step": 16994
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655086104295904e-06,
+      "loss": 0.4708,
+      "step": 16995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653996650065638e-06,
+      "loss": 0.4442,
+      "step": 16996
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.652907225301303e-06,
+      "loss": 0.4401,
+      "step": 16997
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651817830016587e-06,
+      "loss": 0.4627,
+      "step": 16998
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65072846422517e-06,
+      "loss": 0.4209,
+      "step": 16999
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649639127940734e-06,
+      "loss": 0.4724,
+      "step": 17000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648549821176964e-06,
+      "loss": 0.4639,
+      "step": 17001
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.64746054394754e-06,
+      "loss": 0.4054,
+      "step": 17002
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646371296266139e-06,
+      "loss": 0.4316,
+      "step": 17003
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.645282078146452e-06,
+      "loss": 0.4291,
+      "step": 17004
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.64419288960215e-06,
+      "loss": 0.4407,
+      "step": 17005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.643103730646918e-06,
+      "loss": 0.3981,
+      "step": 17006
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.64201460129444e-06,
+      "loss": 0.456,
+      "step": 17007
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.640925501558384e-06,
+      "loss": 0.4324,
+      "step": 17008
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639836431452438e-06,
+      "loss": 0.4972,
+      "step": 17009
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63874739099028e-06,
+      "loss": 0.4117,
+      "step": 17010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637658380185585e-06,
+      "loss": 0.4963,
+      "step": 17011
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636569399052035e-06,
+      "loss": 0.4577,
+      "step": 17012
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635480447603308e-06,
+      "loss": 0.4985,
+      "step": 17013
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634391525853073e-06,
+      "loss": 0.4514,
+      "step": 17014
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.633302633815014e-06,
+      "loss": 0.4329,
+      "step": 17015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632213771502811e-06,
+      "loss": 0.4368,
+      "step": 17016
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631124938930133e-06,
+      "loss": 0.4879,
+      "step": 17017
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.630036136110656e-06,
+      "loss": 0.4158,
+      "step": 17018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628947363058059e-06,
+      "loss": 0.4824,
+      "step": 17019
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62785861978602e-06,
+      "loss": 0.4624,
+      "step": 17020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626769906308204e-06,
+      "loss": 0.4634,
+      "step": 17021
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625681222638295e-06,
+      "loss": 0.4622,
+      "step": 17022
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624592568789963e-06,
+      "loss": 0.4497,
+      "step": 17023
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623503944776879e-06,
+      "loss": 0.4331,
+      "step": 17024
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6224153506127174e-06,
+      "loss": 0.494,
+      "step": 17025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6213267863111564e-06,
+      "loss": 0.4342,
+      "step": 17026
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620238251885863e-06,
+      "loss": 0.4551,
+      "step": 17027
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619149747350507e-06,
+      "loss": 0.4622,
+      "step": 17028
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6180612727187685e-06,
+      "loss": 0.4281,
+      "step": 17029
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61697282800431e-06,
+      "loss": 0.453,
+      "step": 17030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615884413220804e-06,
+      "loss": 0.4464,
+      "step": 17031
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61479602838193e-06,
+      "loss": 0.4115,
+      "step": 17032
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6137076735013485e-06,
+      "loss": 0.4261,
+      "step": 17033
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61261934859273e-06,
+      "loss": 0.488,
+      "step": 17034
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61153105366975e-06,
+      "loss": 0.451,
+      "step": 17035
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610442788746071e-06,
+      "loss": 0.4601,
+      "step": 17036
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609354553835365e-06,
+      "loss": 0.4197,
+      "step": 17037
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608266348951301e-06,
+      "loss": 0.4389,
+      "step": 17038
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.607178174107545e-06,
+      "loss": 0.417,
+      "step": 17039
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606090029317763e-06,
+      "loss": 0.4306,
+      "step": 17040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6050019145956285e-06,
+      "loss": 0.4464,
+      "step": 17041
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6039138299547995e-06,
+      "loss": 0.5069,
+      "step": 17042
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6028257754089495e-06,
+      "loss": 0.4324,
+      "step": 17043
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601737750971744e-06,
+      "loss": 0.3799,
+      "step": 17044
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600649756656843e-06,
+      "loss": 0.3997,
+      "step": 17045
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599561792477914e-06,
+      "loss": 0.4765,
+      "step": 17046
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598473858448629e-06,
+      "loss": 0.4561,
+      "step": 17047
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597385954582643e-06,
+      "loss": 0.4924,
+      "step": 17048
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596298080893625e-06,
+      "loss": 0.3777,
+      "step": 17049
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5952102373952395e-06,
+      "loss": 0.3896,
+      "step": 17050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.594122424101145e-06,
+      "loss": 0.4455,
+      "step": 17051
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593034641025007e-06,
+      "loss": 0.4476,
+      "step": 17052
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5919468881804925e-06,
+      "loss": 0.4903,
+      "step": 17053
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590859165581258e-06,
+      "loss": 0.4319,
+      "step": 17054
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5897714732409655e-06,
+      "loss": 0.4073,
+      "step": 17055
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588683811173279e-06,
+      "loss": 0.4396,
+      "step": 17056
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.587596179391862e-06,
+      "loss": 0.4553,
+      "step": 17057
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5865085779103675e-06,
+      "loss": 0.4249,
+      "step": 17058
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585421006742464e-06,
+      "loss": 0.4312,
+      "step": 17059
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58433346590181e-06,
+      "loss": 0.4438,
+      "step": 17060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5832459554020565e-06,
+      "loss": 0.4624,
+      "step": 17061
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5821584752568715e-06,
+      "loss": 0.4715,
+      "step": 17062
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.581071025479915e-06,
+      "loss": 0.4239,
+      "step": 17063
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.57998360608484e-06,
+      "loss": 0.4741,
+      "step": 17064
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578896217085303e-06,
+      "loss": 0.4449,
+      "step": 17065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577808858494971e-06,
+      "loss": 0.4741,
+      "step": 17066
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.57672153032749e-06,
+      "loss": 0.4298,
+      "step": 17067
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5756342325965226e-06,
+      "loss": 0.4393,
+      "step": 17068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.57454696531573e-06,
+      "loss": 0.4535,
+      "step": 17069
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.573459728498761e-06,
+      "loss": 0.5013,
+      "step": 17070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.57237252215927e-06,
+      "loss": 0.471,
+      "step": 17071
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571285346310923e-06,
+      "loss": 0.4422,
+      "step": 17072
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570198200967363e-06,
+      "loss": 0.4167,
+      "step": 17073
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56911108614225e-06,
+      "loss": 0.4366,
+      "step": 17074
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.568024001849239e-06,
+      "loss": 0.4879,
+      "step": 17075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.566936948101983e-06,
+      "loss": 0.452,
+      "step": 17076
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.565849924914132e-06,
+      "loss": 0.3982,
+      "step": 17077
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564762932299349e-06,
+      "loss": 0.3972,
+      "step": 17078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.563675970271273e-06,
+      "loss": 0.4673,
+      "step": 17079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.562589038843567e-06,
+      "loss": 0.4632,
+      "step": 17080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561502138029881e-06,
+      "loss": 0.4955,
+      "step": 17081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56041526784386e-06,
+      "loss": 0.461,
+      "step": 17082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559328428299161e-06,
+      "loss": 0.4652,
+      "step": 17083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5582416194094365e-06,
+      "loss": 0.4612,
+      "step": 17084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.55715484118833e-06,
+      "loss": 0.4351,
+      "step": 17085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.556068093649499e-06,
+      "loss": 0.494,
+      "step": 17086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.55498137680659e-06,
+      "loss": 0.445,
+      "step": 17087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.553894690673249e-06,
+      "loss": 0.4703,
+      "step": 17088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.552808035263126e-06,
+      "loss": 0.4318,
+      "step": 17089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5517214105898776e-06,
+      "loss": 0.393,
+      "step": 17090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.550634816667142e-06,
+      "loss": 0.4212,
+      "step": 17091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549548253508567e-06,
+      "loss": 0.4695,
+      "step": 17092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.548461721127806e-06,
+      "loss": 0.4337,
+      "step": 17093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.547375219538505e-06,
+      "loss": 0.4589,
+      "step": 17094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546288748754304e-06,
+      "loss": 0.4925,
+      "step": 17095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545202308788858e-06,
+      "loss": 0.461,
+      "step": 17096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54411589965581e-06,
+      "loss": 0.4752,
+      "step": 17097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5430295213687984e-06,
+      "loss": 0.4196,
+      "step": 17098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.541943173941475e-06,
+      "loss": 0.4457,
+      "step": 17099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540856857387486e-06,
+      "loss": 0.4229,
+      "step": 17100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.539770571720471e-06,
+      "loss": 0.4458,
+      "step": 17101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538684316954073e-06,
+      "loss": 0.4959,
+      "step": 17102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.537598093101943e-06,
+      "loss": 0.4174,
+      "step": 17103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5365119001777144e-06,
+      "loss": 0.465,
+      "step": 17104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535425738195034e-06,
+      "loss": 0.496,
+      "step": 17105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534339607167549e-06,
+      "loss": 0.4748,
+      "step": 17106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.533253507108894e-06,
+      "loss": 0.4359,
+      "step": 17107
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.532167438032711e-06,
+      "loss": 0.4134,
+      "step": 17108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5310813999526475e-06,
+      "loss": 0.4703,
+      "step": 17109
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529995392882336e-06,
+      "loss": 0.4499,
+      "step": 17110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528909416835421e-06,
+      "loss": 0.4476,
+      "step": 17111
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527823471825546e-06,
+      "loss": 0.4438,
+      "step": 17112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526737557866343e-06,
+      "loss": 0.485,
+      "step": 17113
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.525651674971454e-06,
+      "loss": 0.3889,
+      "step": 17114
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524565823154523e-06,
+      "loss": 0.4637,
+      "step": 17115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523480002429179e-06,
+      "loss": 0.4599,
+      "step": 17116
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522394212809068e-06,
+      "loss": 0.4563,
+      "step": 17117
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.521308454307825e-06,
+      "loss": 0.4554,
+      "step": 17118
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5202227269390835e-06,
+      "loss": 0.4515,
+      "step": 17119
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5191370307164836e-06,
+      "loss": 0.4708,
+      "step": 17120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.518051365653664e-06,
+      "loss": 0.4642,
+      "step": 17121
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516965731764259e-06,
+      "loss": 0.4476,
+      "step": 17122
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5158801290619e-06,
+      "loss": 0.4663,
+      "step": 17123
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51479455756023e-06,
+      "loss": 0.4495,
+      "step": 17124
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.513709017272875e-06,
+      "loss": 0.4375,
+      "step": 17125
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5126235082134735e-06,
+      "loss": 0.4382,
+      "step": 17126
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.511538030395665e-06,
+      "loss": 0.4619,
+      "step": 17127
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510452583833076e-06,
+      "loss": 0.4009,
+      "step": 17128
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.50936716853934e-06,
+      "loss": 0.4993,
+      "step": 17129
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.508281784528092e-06,
+      "loss": 0.3963,
+      "step": 17130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.507196431812969e-06,
+      "loss": 0.4346,
+      "step": 17131
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506111110407595e-06,
+      "loss": 0.4167,
+      "step": 17132
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.505025820325603e-06,
+      "loss": 0.3996,
+      "step": 17133
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503940561580631e-06,
+      "loss": 0.4111,
+      "step": 17134
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.502855334186302e-06,
+      "loss": 0.4147,
+      "step": 17135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.501770138156249e-06,
+      "loss": 0.4642,
+      "step": 17136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500684973504108e-06,
+      "loss": 0.4608,
+      "step": 17137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499599840243501e-06,
+      "loss": 0.4717,
+      "step": 17138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.498514738388058e-06,
+      "loss": 0.4676,
+      "step": 17139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497429667951415e-06,
+      "loss": 0.4288,
+      "step": 17140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.49634462894719e-06,
+      "loss": 0.4862,
+      "step": 17141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495259621389019e-06,
+      "loss": 0.4326,
+      "step": 17142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4941746452905285e-06,
+      "loss": 0.4695,
+      "step": 17143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493089700665343e-06,
+      "loss": 0.4796,
+      "step": 17144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.492004787527091e-06,
+      "loss": 0.4672,
+      "step": 17145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490919905889403e-06,
+      "loss": 0.418,
+      "step": 17146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.489835055765897e-06,
+      "loss": 0.4845,
+      "step": 17147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4887502371702045e-06,
+      "loss": 0.4502,
+      "step": 17148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487665450115953e-06,
+      "loss": 0.4963,
+      "step": 17149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4865806946167595e-06,
+      "loss": 0.4177,
+      "step": 17150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.485495970686254e-06,
+      "loss": 0.4689,
+      "step": 17151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484411278338064e-06,
+      "loss": 0.4281,
+      "step": 17152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483326617585804e-06,
+      "loss": 0.472,
+      "step": 17153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.482241988443105e-06,
+      "loss": 0.4266,
+      "step": 17154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.48115739092359e-06,
+      "loss": 0.4129,
+      "step": 17155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480072825040875e-06,
+      "loss": 0.4567,
+      "step": 17156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4789882908085865e-06,
+      "loss": 0.4485,
+      "step": 17157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4779037882403505e-06,
+      "loss": 0.4628,
+      "step": 17158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.476819317349781e-06,
+      "loss": 0.4586,
+      "step": 17159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4757348781505025e-06,
+      "loss": 0.4579,
+      "step": 17160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.474650470656137e-06,
+      "loss": 0.4403,
+      "step": 17161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.473566094880301e-06,
+      "loss": 0.502,
+      "step": 17162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.472481750836615e-06,
+      "loss": 0.4835,
+      "step": 17163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471397438538705e-06,
+      "loss": 0.4319,
+      "step": 17164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.470313158000182e-06,
+      "loss": 0.4291,
+      "step": 17165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469228909234665e-06,
+      "loss": 0.4117,
+      "step": 17166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.468144692255774e-06,
+      "loss": 0.4073,
+      "step": 17167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4670605070771326e-06,
+      "loss": 0.4666,
+      "step": 17168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.465976353712352e-06,
+      "loss": 0.4208,
+      "step": 17169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.464892232175047e-06,
+      "loss": 0.4509,
+      "step": 17170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463808142478842e-06,
+      "loss": 0.4356,
+      "step": 17171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.462724084637345e-06,
+      "loss": 0.4132,
+      "step": 17172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461640058664175e-06,
+      "loss": 0.4047,
+      "step": 17173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460556064572953e-06,
+      "loss": 0.45,
+      "step": 17174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4594721023772855e-06,
+      "loss": 0.5067,
+      "step": 17175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4583881720907895e-06,
+      "loss": 0.43,
+      "step": 17176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457304273727084e-06,
+      "loss": 0.4479,
+      "step": 17177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456220407299776e-06,
+      "loss": 0.4895,
+      "step": 17178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4551365728224835e-06,
+      "loss": 0.464,
+      "step": 17179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.454052770308818e-06,
+      "loss": 0.4679,
+      "step": 17180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.452968999772393e-06,
+      "loss": 0.436,
+      "step": 17181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451885261226818e-06,
+      "loss": 0.4818,
+      "step": 17182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450801554685711e-06,
+      "loss": 0.4083,
+      "step": 17183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.449717880162675e-06,
+      "loss": 0.4494,
+      "step": 17184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448634237671326e-06,
+      "loss": 0.4073,
+      "step": 17185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447550627225277e-06,
+      "loss": 0.4608,
+      "step": 17186
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.446467048838131e-06,
+      "loss": 0.4653,
+      "step": 17187
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445383502523502e-06,
+      "loss": 0.4647,
+      "step": 17188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4442999882950025e-06,
+      "loss": 0.4693,
+      "step": 17189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.443216506166235e-06,
+      "loss": 0.4313,
+      "step": 17190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442133056150814e-06,
+      "loss": 0.4729,
+      "step": 17191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.441049638262347e-06,
+      "loss": 0.4411,
+      "step": 17192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.439966252514434e-06,
+      "loss": 0.4778,
+      "step": 17193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43888289892069e-06,
+      "loss": 0.479,
+      "step": 17194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437799577494725e-06,
+      "loss": 0.4227,
+      "step": 17195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4367162882501365e-06,
+      "loss": 0.4309,
+      "step": 17196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.435633031200534e-06,
+      "loss": 0.4657,
+      "step": 17197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4345498063595276e-06,
+      "loss": 0.4712,
+      "step": 17198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433466613740717e-06,
+      "loss": 0.4054,
+      "step": 17199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432383453357708e-06,
+      "loss": 0.4373,
+      "step": 17200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4313003252241085e-06,
+      "loss": 0.4247,
+      "step": 17201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.430217229353525e-06,
+      "loss": 0.4626,
+      "step": 17202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4291341657595504e-06,
+      "loss": 0.4594,
+      "step": 17203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428051134455796e-06,
+      "loss": 0.4478,
+      "step": 17204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4269681354558676e-06,
+      "loss": 0.4672,
+      "step": 17205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4258851687733605e-06,
+      "loss": 0.4425,
+      "step": 17206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424802234421877e-06,
+      "loss": 0.4426,
+      "step": 17207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423719332415028e-06,
+      "loss": 0.485,
+      "step": 17208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422636462766404e-06,
+      "loss": 0.4652,
+      "step": 17209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42155362548961e-06,
+      "loss": 0.4522,
+      "step": 17210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42047082059825e-06,
+      "loss": 0.4413,
+      "step": 17211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41938804810592e-06,
+      "loss": 0.4776,
+      "step": 17212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418305308026218e-06,
+      "loss": 0.4572,
+      "step": 17213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.417222600372751e-06,
+      "loss": 0.4581,
+      "step": 17214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416139925159107e-06,
+      "loss": 0.4474,
+      "step": 17215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.415057282398894e-06,
+      "loss": 0.4151,
+      "step": 17216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413974672105705e-06,
+      "loss": 0.4898,
+      "step": 17217
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4128920942931404e-06,
+      "loss": 0.4045,
+      "step": 17218
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.4331,
+      "step": 17219
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410727036164267e-06,
+      "loss": 0.4407,
+      "step": 17220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.409644555875149e-06,
+      "loss": 0.3739,
+      "step": 17221
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408562108121044e-06,
+      "loss": 0.4879,
+      "step": 17222
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4074796929155455e-06,
+      "loss": 0.4201,
+      "step": 17223
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4063973102722446e-06,
+      "loss": 0.4769,
+      "step": 17224
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405314960204737e-06,
+      "loss": 0.4667,
+      "step": 17225
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404232642726623e-06,
+      "loss": 0.4719,
+      "step": 17226
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.403150357851488e-06,
+      "loss": 0.4212,
+      "step": 17227
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402068105592934e-06,
+      "loss": 0.4384,
+      "step": 17228
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4009858859645515e-06,
+      "loss": 0.4331,
+      "step": 17229
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399903698979927e-06,
+      "loss": 0.462,
+      "step": 17230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398821544652657e-06,
+      "loss": 0.4309,
+      "step": 17231
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.39773942299634e-06,
+      "loss": 0.4054,
+      "step": 17232
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3966573340245575e-06,
+      "loss": 0.4641,
+      "step": 17233
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395575277750904e-06,
+      "loss": 0.5143,
+      "step": 17234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394493254188974e-06,
+      "loss": 0.4861,
+      "step": 17235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393411263352351e-06,
+      "loss": 0.3877,
+      "step": 17236
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3923293052546285e-06,
+      "loss": 0.4226,
+      "step": 17237
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391247379909399e-06,
+      "loss": 0.4279,
+      "step": 17238
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3901654873302495e-06,
+      "loss": 0.4298,
+      "step": 17239
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3890836275307645e-06,
+      "loss": 0.4209,
+      "step": 17240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.388001800524534e-06,
+      "loss": 0.4741,
+      "step": 17241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386920006325151e-06,
+      "loss": 0.486,
+      "step": 17242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385838244946199e-06,
+      "loss": 0.4516,
+      "step": 17243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384756516401261e-06,
+      "loss": 0.4671,
+      "step": 17244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383674820703934e-06,
+      "loss": 0.451,
+      "step": 17245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.382593157867792e-06,
+      "loss": 0.4612,
+      "step": 17246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3815115279064264e-06,
+      "loss": 0.4302,
+      "step": 17247
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380429930833428e-06,
+      "loss": 0.4693,
+      "step": 17248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379348366662373e-06,
+      "loss": 0.4415,
+      "step": 17249
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378266835406849e-06,
+      "loss": 0.3757,
+      "step": 17250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377185337080443e-06,
+      "loss": 0.4445,
+      "step": 17251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376103871696732e-06,
+      "loss": 0.4447,
+      "step": 17252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.375022439269306e-06,
+      "loss": 0.4433,
+      "step": 17253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373941039811747e-06,
+      "loss": 0.4971,
+      "step": 17254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3728596733376315e-06,
+      "loss": 0.4341,
+      "step": 17255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3717783398605446e-06,
+      "loss": 0.4238,
+      "step": 17256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370697039394075e-06,
+      "loss": 0.3986,
+      "step": 17257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3696157719517926e-06,
+      "loss": 0.4257,
+      "step": 17258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368534537547285e-06,
+      "loss": 0.4292,
+      "step": 17259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367453336194134e-06,
+      "loss": 0.4111,
+      "step": 17260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366372167905912e-06,
+      "loss": 0.4589,
+      "step": 17261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365291032696203e-06,
+      "loss": 0.471,
+      "step": 17262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36420993057859e-06,
+      "loss": 0.4012,
+      "step": 17263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363128861566647e-06,
+      "loss": 0.4814,
+      "step": 17264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36204782567395e-06,
+      "loss": 0.4516,
+      "step": 17265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.360966822914085e-06,
+      "loss": 0.413,
+      "step": 17266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35988585330062e-06,
+      "loss": 0.4936,
+      "step": 17267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3588049168471365e-06,
+      "loss": 0.4446,
+      "step": 17268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.357724013567215e-06,
+      "loss": 0.4165,
+      "step": 17269
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.356643143474427e-06,
+      "loss": 0.4199,
+      "step": 17270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355562306582346e-06,
+      "loss": 0.4279,
+      "step": 17271
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354481502904555e-06,
+      "loss": 0.5097,
+      "step": 17272
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35340073245462e-06,
+      "loss": 0.4336,
+      "step": 17273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352319995246122e-06,
+      "loss": 0.4394,
+      "step": 17274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.351239291292631e-06,
+      "loss": 0.4075,
+      "step": 17275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350158620607728e-06,
+      "loss": 0.394,
+      "step": 17276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.349077983204976e-06,
+      "loss": 0.509,
+      "step": 17277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347997379097953e-06,
+      "loss": 0.4153,
+      "step": 17278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3469168083002375e-06,
+      "loss": 0.426,
+      "step": 17279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345836270825391e-06,
+      "loss": 0.4428,
+      "step": 17280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.344755766686988e-06,
+      "loss": 0.4451,
+      "step": 17281
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343675295898607e-06,
+      "loss": 0.4234,
+      "step": 17282
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.342594858473807e-06,
+      "loss": 0.3891,
+      "step": 17283
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.341514454426168e-06,
+      "loss": 0.4604,
+      "step": 17284
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.340434083769254e-06,
+      "loss": 0.4621,
+      "step": 17285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.339353746516639e-06,
+      "loss": 0.4189,
+      "step": 17286
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338273442681887e-06,
+      "loss": 0.4855,
+      "step": 17287
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.337193172278574e-06,
+      "loss": 0.3905,
+      "step": 17288
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336112935320261e-06,
+      "loss": 0.4176,
+      "step": 17289
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3350327318205195e-06,
+      "loss": 0.4236,
+      "step": 17290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.333952561792918e-06,
+      "loss": 0.3722,
+      "step": 17291
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332872425251017e-06,
+      "loss": 0.4055,
+      "step": 17292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3317923222083885e-06,
+      "loss": 0.3999,
+      "step": 17293
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330712252678602e-06,
+      "loss": 0.4795,
+      "step": 17294
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.329632216675216e-06,
+      "loss": 0.4946,
+      "step": 17295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328552214211799e-06,
+      "loss": 0.4099,
+      "step": 17296
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3274722453019195e-06,
+      "loss": 0.3807,
+      "step": 17297
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.326392309959134e-06,
+      "loss": 0.4462,
+      "step": 17298
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.32531240819701e-06,
+      "loss": 0.4937,
+      "step": 17299
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.324232540029117e-06,
+      "loss": 0.4698,
+      "step": 17300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.32315270546901e-06,
+      "loss": 0.4447,
+      "step": 17301
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.322072904530253e-06,
+      "loss": 0.4324,
+      "step": 17302
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320993137226415e-06,
+      "loss": 0.4441,
+      "step": 17303
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.31991340357105e-06,
+      "loss": 0.4748,
+      "step": 17304
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3188337035777215e-06,
+      "loss": 0.452,
+      "step": 17305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317754037259996e-06,
+      "loss": 0.5074,
+      "step": 17306
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316674404631427e-06,
+      "loss": 0.4245,
+      "step": 17307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315594805705577e-06,
+      "loss": 0.4317,
+      "step": 17308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314515240496012e-06,
+      "loss": 0.352,
+      "step": 17309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313435709016279e-06,
+      "loss": 0.4489,
+      "step": 17310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.312356211279947e-06,
+      "loss": 0.4897,
+      "step": 17311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.311276747300571e-06,
+      "loss": 0.4308,
+      "step": 17312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310197317091712e-06,
+      "loss": 0.4593,
+      "step": 17313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309117920666921e-06,
+      "loss": 0.4885,
+      "step": 17314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.308038558039762e-06,
+      "loss": 0.4509,
+      "step": 17315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306959229223791e-06,
+      "loss": 0.4701,
+      "step": 17316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305879934232562e-06,
+      "loss": 0.41,
+      "step": 17317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3048006730796306e-06,
+      "loss": 0.3753,
+      "step": 17318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.303721445778558e-06,
+      "loss": 0.4326,
+      "step": 17319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302642252342889e-06,
+      "loss": 0.4641,
+      "step": 17320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301563092786187e-06,
+      "loss": 0.4321,
+      "step": 17321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300483967122004e-06,
+      "loss": 0.5078,
+      "step": 17322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.299404875363894e-06,
+      "loss": 0.5021,
+      "step": 17323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298325817525407e-06,
+      "loss": 0.4513,
+      "step": 17324
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.297246793620104e-06,
+      "loss": 0.5141,
+      "step": 17325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296167803661528e-06,
+      "loss": 0.4102,
+      "step": 17326
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295088847663238e-06,
+      "loss": 0.3958,
+      "step": 17327
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294009925638787e-06,
+      "loss": 0.4401,
+      "step": 17328
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292931037601716e-06,
+      "loss": 0.4291,
+      "step": 17329
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2918521835655845e-06,
+      "loss": 0.4084,
+      "step": 17330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290773363543946e-06,
+      "loss": 0.4901,
+      "step": 17331
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.28969457755034e-06,
+      "loss": 0.5092,
+      "step": 17332
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2886158255983255e-06,
+      "loss": 0.451,
+      "step": 17333
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.287537107701449e-06,
+      "loss": 0.5378,
+      "step": 17334
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286458423873255e-06,
+      "loss": 0.3939,
+      "step": 17335
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285379774127295e-06,
+      "loss": 0.4318,
+      "step": 17336
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284301158477121e-06,
+      "loss": 0.4995,
+      "step": 17337
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283222576936274e-06,
+      "loss": 0.4702,
+      "step": 17338
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2821440295183025e-06,
+      "loss": 0.4734,
+      "step": 17339
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.281065516236758e-06,
+      "loss": 0.4368,
+      "step": 17340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.279987037105179e-06,
+      "loss": 0.4111,
+      "step": 17341
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278908592137114e-06,
+      "loss": 0.4732,
+      "step": 17342
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.277830181346115e-06,
+      "loss": 0.4725,
+      "step": 17343
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276751804745719e-06,
+      "loss": 0.4452,
+      "step": 17344
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27567346234947e-06,
+      "loss": 0.4582,
+      "step": 17345
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274595154170919e-06,
+      "loss": 0.4426,
+      "step": 17346
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.273516880223601e-06,
+      "loss": 0.4204,
+      "step": 17347
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.272438640521067e-06,
+      "loss": 0.4961,
+      "step": 17348
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271360435076855e-06,
+      "loss": 0.4193,
+      "step": 17349
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270282263904511e-06,
+      "loss": 0.4113,
+      "step": 17350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269204127017572e-06,
+      "loss": 0.4069,
+      "step": 17351
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268126024429581e-06,
+      "loss": 0.4697,
+      "step": 17352
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267047956154085e-06,
+      "loss": 0.4464,
+      "step": 17353
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.265969922204617e-06,
+      "loss": 0.4755,
+      "step": 17354
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.264891922594719e-06,
+      "loss": 0.4015,
+      "step": 17355
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263813957337935e-06,
+      "loss": 0.4199,
+      "step": 17356
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262736026447797e-06,
+      "loss": 0.4661,
+      "step": 17357
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26165812993785e-06,
+      "loss": 0.45,
+      "step": 17358
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260580267821629e-06,
+      "loss": 0.4557,
+      "step": 17359
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259502440112675e-06,
+      "loss": 0.4107,
+      "step": 17360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25842464682452e-06,
+      "loss": 0.3769,
+      "step": 17361
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.257346887970711e-06,
+      "loss": 0.4317,
+      "step": 17362
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256269163564773e-06,
+      "loss": 0.4541,
+      "step": 17363
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.255191473620251e-06,
+      "loss": 0.4164,
+      "step": 17364
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25411381815068e-06,
+      "loss": 0.449,
+      "step": 17365
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2530361971695885e-06,
+      "loss": 0.4378,
+      "step": 17366
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251958610690517e-06,
+      "loss": 0.5089,
+      "step": 17367
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250881058727002e-06,
+      "loss": 0.4236,
+      "step": 17368
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249803541292572e-06,
+      "loss": 0.4523,
+      "step": 17369
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.248726058400766e-06,
+      "loss": 0.4072,
+      "step": 17370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.247648610065115e-06,
+      "loss": 0.4638,
+      "step": 17371
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.246571196299148e-06,
+      "loss": 0.4629,
+      "step": 17372
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.245493817116402e-06,
+      "loss": 0.4518,
+      "step": 17373
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244416472530411e-06,
+      "loss": 0.4373,
+      "step": 17374
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2433391625547025e-06,
+      "loss": 0.3946,
+      "step": 17375
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.242261887202806e-06,
+      "loss": 0.4506,
+      "step": 17376
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241184646488258e-06,
+      "loss": 0.4305,
+      "step": 17377
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240107440424582e-06,
+      "loss": 0.487,
+      "step": 17378
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239030269025311e-06,
+      "loss": 0.4197,
+      "step": 17379
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.237953132303979e-06,
+      "loss": 0.4946,
+      "step": 17380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236876030274107e-06,
+      "loss": 0.4583,
+      "step": 17381
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2357989629492254e-06,
+      "loss": 0.4073,
+      "step": 17382
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23472193034287e-06,
+      "loss": 0.4498,
+      "step": 17383
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.233644932468555e-06,
+      "loss": 0.4754,
+      "step": 17384
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232567969339816e-06,
+      "loss": 0.4292,
+      "step": 17385
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.231491040970179e-06,
+      "loss": 0.4402,
+      "step": 17386
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230414147373172e-06,
+      "loss": 0.4093,
+      "step": 17387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229337288562315e-06,
+      "loss": 0.4242,
+      "step": 17388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228260464551138e-06,
+      "loss": 0.4046,
+      "step": 17389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227183675353166e-06,
+      "loss": 0.4587,
+      "step": 17390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.226106920981921e-06,
+      "loss": 0.4568,
+      "step": 17391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.225030201450927e-06,
+      "loss": 0.4719,
+      "step": 17392
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.223953516773714e-06,
+      "loss": 0.4499,
+      "step": 17393
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.222876866963795e-06,
+      "loss": 0.4418,
+      "step": 17394
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221800252034701e-06,
+      "loss": 0.4257,
+      "step": 17395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220723671999954e-06,
+      "loss": 0.4348,
+      "step": 17396
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2196471268730686e-06,
+      "loss": 0.4794,
+      "step": 17397
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.21857061666757e-06,
+      "loss": 0.4928,
+      "step": 17398
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2174941413969855e-06,
+      "loss": 0.4753,
+      "step": 17399
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216417701074826e-06,
+      "loss": 0.4538,
+      "step": 17400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215341295714618e-06,
+      "loss": 0.4378,
+      "step": 17401
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2142649253298815e-06,
+      "loss": 0.4597,
+      "step": 17402
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213188589934129e-06,
+      "loss": 0.4528,
+      "step": 17403
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212112289540885e-06,
+      "loss": 0.413,
+      "step": 17404
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.211036024163669e-06,
+      "loss": 0.4363,
+      "step": 17405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209959793815997e-06,
+      "loss": 0.4409,
+      "step": 17406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208883598511382e-06,
+      "loss": 0.4829,
+      "step": 17407
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20780743826335e-06,
+      "loss": 0.4965,
+      "step": 17408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20673131308541e-06,
+      "loss": 0.4438,
+      "step": 17409
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20565522299108e-06,
+      "loss": 0.45,
+      "step": 17410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204579167993881e-06,
+      "loss": 0.4867,
+      "step": 17411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.203503148107324e-06,
+      "loss": 0.4152,
+      "step": 17412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2024271633449205e-06,
+      "loss": 0.4387,
+      "step": 17413
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2013512137201925e-06,
+      "loss": 0.4094,
+      "step": 17414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2002752992466465e-06,
+      "loss": 0.386,
+      "step": 17415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199199419937801e-06,
+      "loss": 0.4502,
+      "step": 17416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.198123575807168e-06,
+      "loss": 0.4591,
+      "step": 17417
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197047766868259e-06,
+      "loss": 0.3563,
+      "step": 17418
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.195971993134585e-06,
+      "loss": 0.4339,
+      "step": 17419
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.194896254619661e-06,
+      "loss": 0.4529,
+      "step": 17420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193820551337001e-06,
+      "loss": 0.4774,
+      "step": 17421
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192744883300109e-06,
+      "loss": 0.4477,
+      "step": 17422
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1916692505224975e-06,
+      "loss": 0.4264,
+      "step": 17423
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190593653017681e-06,
+      "loss": 0.4742,
+      "step": 17424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.189518090799163e-06,
+      "loss": 0.4154,
+      "step": 17425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.188442563880455e-06,
+      "loss": 0.4645,
+      "step": 17426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.187367072275066e-06,
+      "loss": 0.404,
+      "step": 17427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186291615996505e-06,
+      "loss": 0.4307,
+      "step": 17428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185216195058277e-06,
+      "loss": 0.4857,
+      "step": 17429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.184140809473896e-06,
+      "loss": 0.4114,
+      "step": 17430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183065459256859e-06,
+      "loss": 0.4448,
+      "step": 17431
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18199014442068e-06,
+      "loss": 0.4445,
+      "step": 17432
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180914864978864e-06,
+      "loss": 0.4295,
+      "step": 17433
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179839620944911e-06,
+      "loss": 0.4646,
+      "step": 17434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178764412332331e-06,
+      "loss": 0.4534,
+      "step": 17435
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.177689239154633e-06,
+      "loss": 0.4782,
+      "step": 17436
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17661410142531e-06,
+      "loss": 0.4061,
+      "step": 17437
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.175538999157876e-06,
+      "loss": 0.4448,
+      "step": 17438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1744639323658315e-06,
+      "loss": 0.4793,
+      "step": 17439
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173388901062675e-06,
+      "loss": 0.4828,
+      "step": 17440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.172313905261909e-06,
+      "loss": 0.4593,
+      "step": 17441
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171238944977046e-06,
+      "loss": 0.4418,
+      "step": 17442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170164020221577e-06,
+      "loss": 0.4289,
+      "step": 17443
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169089131009006e-06,
+      "loss": 0.4455,
+      "step": 17444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168014277352837e-06,
+      "loss": 0.414,
+      "step": 17445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166939459266562e-06,
+      "loss": 0.4589,
+      "step": 17446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.165864676763687e-06,
+      "loss": 0.3697,
+      "step": 17447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.164789929857716e-06,
+      "loss": 0.4727,
+      "step": 17448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163715218562138e-06,
+      "loss": 0.4461,
+      "step": 17449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.162640542890456e-06,
+      "loss": 0.4385,
+      "step": 17450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.161565902856169e-06,
+      "loss": 0.4413,
+      "step": 17451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1604912984727705e-06,
+      "loss": 0.4903,
+      "step": 17452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159416729753763e-06,
+      "loss": 0.4035,
+      "step": 17453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15834219671264e-06,
+      "loss": 0.4319,
+      "step": 17454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157267699362898e-06,
+      "loss": 0.4856,
+      "step": 17455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15619323771803e-06,
+      "loss": 0.4737,
+      "step": 17456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1551188117915346e-06,
+      "loss": 0.4566,
+      "step": 17457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.154044421596911e-06,
+      "loss": 0.4702,
+      "step": 17458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152970067147646e-06,
+      "loss": 0.3764,
+      "step": 17459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151895748457236e-06,
+      "loss": 0.3811,
+      "step": 17460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.150821465539178e-06,
+      "loss": 0.4996,
+      "step": 17461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149747218406957e-06,
+      "loss": 0.4675,
+      "step": 17462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.148673007074074e-06,
+      "loss": 0.4658,
+      "step": 17463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.147598831554017e-06,
+      "loss": 0.4422,
+      "step": 17464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.14652469186028e-06,
+      "loss": 0.4617,
+      "step": 17465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145450588006347e-06,
+      "loss": 0.4256,
+      "step": 17466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.144376520005721e-06,
+      "loss": 0.4309,
+      "step": 17467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143302487871882e-06,
+      "loss": 0.4029,
+      "step": 17468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142228491618326e-06,
+      "loss": 0.4894,
+      "step": 17469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1411545312585405e-06,
+      "loss": 0.444,
+      "step": 17470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.140080606806013e-06,
+      "loss": 0.4286,
+      "step": 17471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139006718274231e-06,
+      "loss": 0.461,
+      "step": 17472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137932865676689e-06,
+      "loss": 0.421,
+      "step": 17473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136859049026866e-06,
+      "loss": 0.4023,
+      "step": 17474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1357852683382565e-06,
+      "loss": 0.4453,
+      "step": 17475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.134711523624347e-06,
+      "loss": 0.4372,
+      "step": 17476
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133637814898616e-06,
+      "loss": 0.4411,
+      "step": 17477
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132564142174556e-06,
+      "loss": 0.4494,
+      "step": 17478
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131490505465655e-06,
+      "loss": 0.4297,
+      "step": 17479
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130416904785391e-06,
+      "loss": 0.4779,
+      "step": 17480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.129343340147249e-06,
+      "loss": 0.4412,
+      "step": 17481
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12826981156472e-06,
+      "loss": 0.4797,
+      "step": 17482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.127196319051279e-06,
+      "loss": 0.4538,
+      "step": 17483
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.126122862620412e-06,
+      "loss": 0.4012,
+      "step": 17484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1250494422856085e-06,
+      "loss": 0.3825,
+      "step": 17485
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123976058060343e-06,
+      "loss": 0.4382,
+      "step": 17486
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122902709958096e-06,
+      "loss": 0.4456,
+      "step": 17487
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.121829397992357e-06,
+      "loss": 0.4778,
+      "step": 17488
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120756122176597e-06,
+      "loss": 0.4468,
+      "step": 17489
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119682882524306e-06,
+      "loss": 0.4662,
+      "step": 17490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.118609679048957e-06,
+      "loss": 0.4441,
+      "step": 17491
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117536511764031e-06,
+      "loss": 0.4095,
+      "step": 17492
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116463380683008e-06,
+      "loss": 0.4374,
+      "step": 17493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.115390285819365e-06,
+      "loss": 0.4141,
+      "step": 17494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.114317227186587e-06,
+      "loss": 0.441,
+      "step": 17495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113244204798142e-06,
+      "loss": 0.4265,
+      "step": 17496
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.112171218667512e-06,
+      "loss": 0.4264,
+      "step": 17497
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111098268808177e-06,
+      "loss": 0.4764,
+      "step": 17498
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110025355233603e-06,
+      "loss": 0.4618,
+      "step": 17499
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108952477957276e-06,
+      "loss": 0.4817,
+      "step": 17500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107879636992667e-06,
+      "loss": 0.4593,
+      "step": 17501
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106806832353254e-06,
+      "loss": 0.4765,
+      "step": 17502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.105734064052505e-06,
+      "loss": 0.4346,
+      "step": 17503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.104661332103904e-06,
+      "loss": 0.4462,
+      "step": 17504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1035886365209125e-06,
+      "loss": 0.4486,
+      "step": 17505
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102515977317014e-06,
+      "loss": 0.4774,
+      "step": 17506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1014433545056785e-06,
+      "loss": 0.4538,
+      "step": 17507
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100370768100372e-06,
+      "loss": 0.4333,
+      "step": 17508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.099298218114572e-06,
+      "loss": 0.4478,
+      "step": 17509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.098225704561752e-06,
+      "loss": 0.4771,
+      "step": 17510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 0.4103,
+      "step": 17511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096080786808921e-06,
+      "loss": 0.484,
+      "step": 17512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.095008382635855e-06,
+      "loss": 0.4756,
+      "step": 17513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.093936014949643e-06,
+      "loss": 0.4429,
+      "step": 17514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092863683763757e-06,
+      "loss": 0.367,
+      "step": 17515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091791389091669e-06,
+      "loss": 0.4655,
+      "step": 17516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090719130946843e-06,
+      "loss": 0.4148,
+      "step": 17517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0896469093427465e-06,
+      "loss": 0.4433,
+      "step": 17518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.088574724292852e-06,
+      "loss": 0.4419,
+      "step": 17519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0875025758106165e-06,
+      "loss": 0.4837,
+      "step": 17520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086430463909512e-06,
+      "loss": 0.396,
+      "step": 17521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.085358388603007e-06,
+      "loss": 0.3841,
+      "step": 17522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084286349904563e-06,
+      "loss": 0.4367,
+      "step": 17523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083214347827645e-06,
+      "loss": 0.4538,
+      "step": 17524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082142382385722e-06,
+      "loss": 0.4744,
+      "step": 17525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.081070453592248e-06,
+      "loss": 0.4291,
+      "step": 17526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0799985614606954e-06,
+      "loss": 0.4033,
+      "step": 17527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078926706004528e-06,
+      "loss": 0.468,
+      "step": 17528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077854887237199e-06,
+      "loss": 0.4693,
+      "step": 17529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.076783105172176e-06,
+      "loss": 0.4907,
+      "step": 17530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.075711359822924e-06,
+      "loss": 0.4681,
+      "step": 17531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074639651202902e-06,
+      "loss": 0.4617,
+      "step": 17532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0735679793255685e-06,
+      "loss": 0.4523,
+      "step": 17533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072496344204382e-06,
+      "loss": 0.3965,
+      "step": 17534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071424745852811e-06,
+      "loss": 0.4709,
+      "step": 17535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.070353184284305e-06,
+      "loss": 0.4511,
+      "step": 17536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069281659512328e-06,
+      "loss": 0.3945,
+      "step": 17537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.068210171550341e-06,
+      "loss": 0.4893,
+      "step": 17538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.067138720411795e-06,
+      "loss": 0.4748,
+      "step": 17539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.06606730611015e-06,
+      "loss": 0.4578,
+      "step": 17540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.064995928658867e-06,
+      "loss": 0.4739,
+      "step": 17541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063924588071396e-06,
+      "loss": 0.4288,
+      "step": 17542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.062853284361199e-06,
+      "loss": 0.4577,
+      "step": 17543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.06178201754173e-06,
+      "loss": 0.4213,
+      "step": 17544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060710787626441e-06,
+      "loss": 0.4283,
+      "step": 17545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0596395946287885e-06,
+      "loss": 0.428,
+      "step": 17546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.058568438562232e-06,
+      "loss": 0.4797,
+      "step": 17547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057497319440218e-06,
+      "loss": 0.406,
+      "step": 17548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056426237276201e-06,
+      "loss": 0.4837,
+      "step": 17549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05535519208364e-06,
+      "loss": 0.4853,
+      "step": 17550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.054284183875977e-06,
+      "loss": 0.432,
+      "step": 17551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05321321266667e-06,
+      "loss": 0.3893,
+      "step": 17552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052142278469174e-06,
+      "loss": 0.4771,
+      "step": 17553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0510713812969346e-06,
+      "loss": 0.4066,
+      "step": 17554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050000521163402e-06,
+      "loss": 0.4559,
+      "step": 17555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048929698082031e-06,
+      "loss": 0.4572,
+      "step": 17556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.047858912066266e-06,
+      "loss": 0.4607,
+      "step": 17557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046788163129558e-06,
+      "loss": 0.471,
+      "step": 17558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.045717451285359e-06,
+      "loss": 0.4654,
+      "step": 17559
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.044646776547112e-06,
+      "loss": 0.4656,
+      "step": 17560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043576138928265e-06,
+      "loss": 0.4351,
+      "step": 17561
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042505538442273e-06,
+      "loss": 0.4476,
+      "step": 17562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041434975102573e-06,
+      "loss": 0.4579,
+      "step": 17563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0403644489226156e-06,
+      "loss": 0.4181,
+      "step": 17564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.03929395991585e-06,
+      "loss": 0.4262,
+      "step": 17565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.038223508095712e-06,
+      "loss": 0.5001,
+      "step": 17566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.037153093475655e-06,
+      "loss": 0.4639,
+      "step": 17567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036082716069121e-06,
+      "loss": 0.4587,
+      "step": 17568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.035012375889556e-06,
+      "loss": 0.4837,
+      "step": 17569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0339420729504005e-06,
+      "loss": 0.4612,
+      "step": 17570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.032871807265097e-06,
+      "loss": 0.4317,
+      "step": 17571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031801578847094e-06,
+      "loss": 0.4178,
+      "step": 17572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0307313877098245e-06,
+      "loss": 0.3936,
+      "step": 17573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.029661233866738e-06,
+      "loss": 0.4492,
+      "step": 17574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.028591117331276e-06,
+      "loss": 0.4813,
+      "step": 17575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027521038116869e-06,
+      "loss": 0.4535,
+      "step": 17576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026450996236967e-06,
+      "loss": 0.4928,
+      "step": 17577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.025380991705008e-06,
+      "loss": 0.4217,
+      "step": 17578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0243110245344295e-06,
+      "loss": 0.4274,
+      "step": 17579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0232410947386705e-06,
+      "loss": 0.3919,
+      "step": 17580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022171202331174e-06,
+      "loss": 0.3876,
+      "step": 17581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.021101347325369e-06,
+      "loss": 0.4445,
+      "step": 17582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.020031529734699e-06,
+      "loss": 0.3988,
+      "step": 17583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018961749572604e-06,
+      "loss": 0.4391,
+      "step": 17584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0178920068525134e-06,
+      "loss": 0.4715,
+      "step": 17585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016822301587864e-06,
+      "loss": 0.4673,
+      "step": 17586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.015752633792098e-06,
+      "loss": 0.4429,
+      "step": 17587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014683003478642e-06,
+      "loss": 0.4371,
+      "step": 17588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.013613410660934e-06,
+      "loss": 0.4448,
+      "step": 17589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.012543855352414e-06,
+      "loss": 0.4901,
+      "step": 17590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0114743375665084e-06,
+      "loss": 0.4365,
+      "step": 17591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01040485731665e-06,
+      "loss": 0.4514,
+      "step": 17592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.009335414616278e-06,
+      "loss": 0.4612,
+      "step": 17593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008266009478817e-06,
+      "loss": 0.4427,
+      "step": 17594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.007196641917704e-06,
+      "loss": 0.4222,
+      "step": 17595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006127311946369e-06,
+      "loss": 0.463,
+      "step": 17596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.005058019578243e-06,
+      "loss": 0.46,
+      "step": 17597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003988764826754e-06,
+      "loss": 0.4692,
+      "step": 17598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002919547705337e-06,
+      "loss": 0.4565,
+      "step": 17599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001850368227414e-06,
+      "loss": 0.4157,
+      "step": 17600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.000781226406422e-06,
+      "loss": 0.4152,
+      "step": 17601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999712122255788e-06,
+      "loss": 0.3858,
+      "step": 17602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998643055788932e-06,
+      "loss": 0.4336,
+      "step": 17603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.997574027019288e-06,
+      "loss": 0.4657,
+      "step": 17604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.996505035960283e-06,
+      "loss": 0.4471,
+      "step": 17605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.995436082625347e-06,
+      "loss": 0.4793,
+      "step": 17606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994367167027898e-06,
+      "loss": 0.4314,
+      "step": 17607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993298289181365e-06,
+      "loss": 0.4155,
+      "step": 17608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992229449099179e-06,
+      "loss": 0.4164,
+      "step": 17609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.991160646794755e-06,
+      "loss": 0.4576,
+      "step": 17610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.990091882281523e-06,
+      "loss": 0.4263,
+      "step": 17611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989023155572908e-06,
+      "loss": 0.4295,
+      "step": 17612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987954466682327e-06,
+      "loss": 0.4631,
+      "step": 17613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986885815623205e-06,
+      "loss": 0.4195,
+      "step": 17614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.985817202408973e-06,
+      "loss": 0.4405,
+      "step": 17615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.984748627053039e-06,
+      "loss": 0.4898,
+      "step": 17616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983680089568835e-06,
+      "loss": 0.436,
+      "step": 17617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9826115899697785e-06,
+      "loss": 0.4106,
+      "step": 17618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.981543128269286e-06,
+      "loss": 0.3886,
+      "step": 17619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.980474704480782e-06,
+      "loss": 0.4559,
+      "step": 17620
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.979406318617688e-06,
+      "loss": 0.4769,
+      "step": 17621
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978337970693418e-06,
+      "loss": 0.4382,
+      "step": 17622
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977269660721389e-06,
+      "loss": 0.4505,
+      "step": 17623
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976201388715027e-06,
+      "loss": 0.4495,
+      "step": 17624
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975133154687741e-06,
+      "loss": 0.4714,
+      "step": 17625
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974064958652953e-06,
+      "loss": 0.3976,
+      "step": 17626
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.972996800624081e-06,
+      "loss": 0.4276,
+      "step": 17627
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9719286806145385e-06,
+      "loss": 0.4993,
+      "step": 17628
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970860598637738e-06,
+      "loss": 0.4811,
+      "step": 17629
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.969792554707102e-06,
+      "loss": 0.4362,
+      "step": 17630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.968724548836038e-06,
+      "loss": 0.4348,
+      "step": 17631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.967656581037965e-06,
+      "loss": 0.4277,
+      "step": 17632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966588651326295e-06,
+      "loss": 0.4805,
+      "step": 17633
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965520759714441e-06,
+      "loss": 0.4749,
+      "step": 17634
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964452906215815e-06,
+      "loss": 0.4139,
+      "step": 17635
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9633850908438346e-06,
+      "loss": 0.4527,
+      "step": 17636
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.962317313611904e-06,
+      "loss": 0.393,
+      "step": 17637
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961249574533439e-06,
+      "loss": 0.3934,
+      "step": 17638
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.960181873621847e-06,
+      "loss": 0.4137,
+      "step": 17639
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.959114210890547e-06,
+      "loss": 0.4663,
+      "step": 17640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958046586352938e-06,
+      "loss": 0.4266,
+      "step": 17641
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956979000022435e-06,
+      "loss": 0.4102,
+      "step": 17642
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.955911451912449e-06,
+      "loss": 0.4784,
+      "step": 17643
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954843942036382e-06,
+      "loss": 0.4097,
+      "step": 17644
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9537764704076446e-06,
+      "loss": 0.4711,
+      "step": 17645
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952709037039651e-06,
+      "loss": 0.4416,
+      "step": 17646
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951641641945796e-06,
+      "loss": 0.4648,
+      "step": 17647
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950574285139498e-06,
+      "loss": 0.3843,
+      "step": 17648
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9495069666341585e-06,
+      "loss": 0.4645,
+      "step": 17649
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.948439686443178e-06,
+      "loss": 0.4674,
+      "step": 17650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947372444579965e-06,
+      "loss": 0.486,
+      "step": 17651
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94630524105793e-06,
+      "loss": 0.426,
+      "step": 17652
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945238075890471e-06,
+      "loss": 0.4549,
+      "step": 17653
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944170949090989e-06,
+      "loss": 0.4152,
+      "step": 17654
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.943103860672896e-06,
+      "loss": 0.4679,
+      "step": 17655
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942036810649586e-06,
+      "loss": 0.506,
+      "step": 17656
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940969799034465e-06,
+      "loss": 0.4544,
+      "step": 17657
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.939902825840938e-06,
+      "loss": 0.4292,
+      "step": 17658
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938835891082401e-06,
+      "loss": 0.4763,
+      "step": 17659
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937768994772256e-06,
+      "loss": 0.4402,
+      "step": 17660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936702136923906e-06,
+      "loss": 0.4053,
+      "step": 17661
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935635317550746e-06,
+      "loss": 0.4087,
+      "step": 17662
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.934568536666179e-06,
+      "loss": 0.4749,
+      "step": 17663
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933501794283604e-06,
+      "loss": 0.4842,
+      "step": 17664
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.932435090416417e-06,
+      "loss": 0.4902,
+      "step": 17665
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9313684250780155e-06,
+      "loss": 0.4236,
+      "step": 17666
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930301798281803e-06,
+      "loss": 0.4277,
+      "step": 17667
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.929235210041168e-06,
+      "loss": 0.4036,
+      "step": 17668
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928168660369512e-06,
+      "loss": 0.4132,
+      "step": 17669
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.927102149280232e-06,
+      "loss": 0.4861,
+      "step": 17670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926035676786716e-06,
+      "loss": 0.4307,
+      "step": 17671
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.924969242902366e-06,
+      "loss": 0.5265,
+      "step": 17672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.923902847640578e-06,
+      "loss": 0.438,
+      "step": 17673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922836491014737e-06,
+      "loss": 0.452,
+      "step": 17674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921770173038245e-06,
+      "loss": 0.4344,
+      "step": 17675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.920703893724491e-06,
+      "loss": 0.3666,
+      "step": 17676
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919637653086872e-06,
+      "loss": 0.4317,
+      "step": 17677
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.918571451138771e-06,
+      "loss": 0.4339,
+      "step": 17678
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917505287893589e-06,
+      "loss": 0.4426,
+      "step": 17679
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916439163364714e-06,
+      "loss": 0.4491,
+      "step": 17680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.915373077565533e-06,
+      "loss": 0.4279,
+      "step": 17681
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914307030509438e-06,
+      "loss": 0.4571,
+      "step": 17682
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9132410222098246e-06,
+      "loss": 0.4504,
+      "step": 17683
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912175052680073e-06,
+      "loss": 0.4401,
+      "step": 17684
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.911109121933576e-06,
+      "loss": 0.4568,
+      "step": 17685
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.910043229983725e-06,
+      "loss": 0.451,
+      "step": 17686
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9089773768438984e-06,
+      "loss": 0.4294,
+      "step": 17687
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9079115625274895e-06,
+      "loss": 0.4356,
+      "step": 17688
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9068457870478886e-06,
+      "loss": 0.445,
+      "step": 17689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.905780050418478e-06,
+      "loss": 0.5007,
+      "step": 17690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9047143526526394e-06,
+      "loss": 0.4798,
+      "step": 17691
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.903648693763765e-06,
+      "loss": 0.4161,
+      "step": 17692
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902583073765235e-06,
+      "loss": 0.4465,
+      "step": 17693
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901517492670433e-06,
+      "loss": 0.4214,
+      "step": 17694
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900451950492748e-06,
+      "loss": 0.4331,
+      "step": 17695
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89938644724556e-06,
+      "loss": 0.4127,
+      "step": 17696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89832098294225e-06,
+      "loss": 0.4439,
+      "step": 17697
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897255557596205e-06,
+      "loss": 0.4616,
+      "step": 17698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8961901712208005e-06,
+      "loss": 0.438,
+      "step": 17699
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895124823829423e-06,
+      "loss": 0.3947,
+      "step": 17700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8940595154354516e-06,
+      "loss": 0.4637,
+      "step": 17701
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.892994246052267e-06,
+      "loss": 0.4576,
+      "step": 17702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891929015693247e-06,
+      "loss": 0.4689,
+      "step": 17703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890863824371776e-06,
+      "loss": 0.4522,
+      "step": 17704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.889798672101226e-06,
+      "loss": 0.3952,
+      "step": 17705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.88873355889498e-06,
+      "loss": 0.487,
+      "step": 17706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.887668484766419e-06,
+      "loss": 0.4169,
+      "step": 17707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8866034497289095e-06,
+      "loss": 0.4624,
+      "step": 17708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.885538453795836e-06,
+      "loss": 0.4684,
+      "step": 17709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884473496980577e-06,
+      "loss": 0.4541,
+      "step": 17710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883408579296503e-06,
+      "loss": 0.4482,
+      "step": 17711
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882343700756992e-06,
+      "loss": 0.4515,
+      "step": 17712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8812788613754185e-06,
+      "loss": 0.4536,
+      "step": 17713
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880214061165161e-06,
+      "loss": 0.4712,
+      "step": 17714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.879149300139585e-06,
+      "loss": 0.4449,
+      "step": 17715
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878084578312071e-06,
+      "loss": 0.4248,
+      "step": 17716
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877019895695991e-06,
+      "loss": 0.4458,
+      "step": 17717
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875955252304712e-06,
+      "loss": 0.4099,
+      "step": 17718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874890648151611e-06,
+      "loss": 0.4243,
+      "step": 17719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.873826083250061e-06,
+      "loss": 0.4559,
+      "step": 17720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872761557613428e-06,
+      "loss": 0.442,
+      "step": 17721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8716970712550855e-06,
+      "loss": 0.4415,
+      "step": 17722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870632624188407e-06,
+      "loss": 0.4649,
+      "step": 17723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8695682164267515e-06,
+      "loss": 0.4496,
+      "step": 17724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868503847983496e-06,
+      "loss": 0.4525,
+      "step": 17725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.867439518872011e-06,
+      "loss": 0.4129,
+      "step": 17726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.86637522910566e-06,
+      "loss": 0.385,
+      "step": 17727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.865310978697808e-06,
+      "loss": 0.5009,
+      "step": 17728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864246767661831e-06,
+      "loss": 0.437,
+      "step": 17729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8631825960110866e-06,
+      "loss": 0.4902,
+      "step": 17730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862118463758944e-06,
+      "loss": 0.4135,
+      "step": 17731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861054370918773e-06,
+      "loss": 0.4674,
+      "step": 17732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859990317503933e-06,
+      "loss": 0.42,
+      "step": 17733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858926303527788e-06,
+      "loss": 0.4487,
+      "step": 17734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857862329003708e-06,
+      "loss": 0.4371,
+      "step": 17735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85679839394505e-06,
+      "loss": 0.3827,
+      "step": 17736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.855734498365183e-06,
+      "loss": 0.4762,
+      "step": 17737
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854670642277466e-06,
+      "loss": 0.4014,
+      "step": 17738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853606825695263e-06,
+      "loss": 0.4348,
+      "step": 17739
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85254304863193e-06,
+      "loss": 0.4251,
+      "step": 17740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851479311100838e-06,
+      "loss": 0.4126,
+      "step": 17741
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850415613115339e-06,
+      "loss": 0.4773,
+      "step": 17742
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849351954688796e-06,
+      "loss": 0.4237,
+      "step": 17743
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.848288335834574e-06,
+      "loss": 0.4414,
+      "step": 17744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.847224756566021e-06,
+      "loss": 0.4141,
+      "step": 17745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846161216896502e-06,
+      "loss": 0.4076,
+      "step": 17746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84509771683938e-06,
+      "loss": 0.4293,
+      "step": 17747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844034256408001e-06,
+      "loss": 0.4871,
+      "step": 17748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.842970835615732e-06,
+      "loss": 0.4488,
+      "step": 17749
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841907454475923e-06,
+      "loss": 0.4264,
+      "step": 17750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840844113001939e-06,
+      "loss": 0.3908,
+      "step": 17751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.839780811207124e-06,
+      "loss": 0.4573,
+      "step": 17752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838717549104843e-06,
+      "loss": 0.3858,
+      "step": 17753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837654326708447e-06,
+      "loss": 0.443,
+      "step": 17754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.836591144031286e-06,
+      "loss": 0.4631,
+      "step": 17755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835528001086718e-06,
+      "loss": 0.4117,
+      "step": 17756
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834464897888099e-06,
+      "loss": 0.4541,
+      "step": 17757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833401834448774e-06,
+      "loss": 0.4387,
+      "step": 17758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832338810782102e-06,
+      "loss": 0.465,
+      "step": 17759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831275826901433e-06,
+      "loss": 0.4392,
+      "step": 17760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830212882820115e-06,
+      "loss": 0.3995,
+      "step": 17761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.829149978551499e-06,
+      "loss": 0.4102,
+      "step": 17762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8280871141089415e-06,
+      "loss": 0.4494,
+      "step": 17763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.827024289505786e-06,
+      "loss": 0.4555,
+      "step": 17764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825961504755379e-06,
+      "loss": 0.3885,
+      "step": 17765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.824898759871079e-06,
+      "loss": 0.517,
+      "step": 17766
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8238360548662245e-06,
+      "loss": 0.4558,
+      "step": 17767
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.822773389754167e-06,
+      "loss": 0.4617,
+      "step": 17768
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821710764548257e-06,
+      "loss": 0.3718,
+      "step": 17769
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8206481792618365e-06,
+      "loss": 0.4597,
+      "step": 17770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.81958563390825e-06,
+      "loss": 0.4941,
+      "step": 17771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8185231285008515e-06,
+      "loss": 0.4432,
+      "step": 17772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.817460663052975e-06,
+      "loss": 0.4592,
+      "step": 17773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8163982375779746e-06,
+      "loss": 0.4496,
+      "step": 17774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.815335852089191e-06,
+      "loss": 0.4202,
+      "step": 17775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814273506599964e-06,
+      "loss": 0.4708,
+      "step": 17776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.813211201123639e-06,
+      "loss": 0.4626,
+      "step": 17777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.812148935673566e-06,
+      "loss": 0.4742,
+      "step": 17778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.811086710263078e-06,
+      "loss": 0.4186,
+      "step": 17779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8100245249055185e-06,
+      "loss": 0.4504,
+      "step": 17780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808962379614232e-06,
+      "loss": 0.4493,
+      "step": 17781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.807900274402554e-06,
+      "loss": 0.3694,
+      "step": 17782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806838209283826e-06,
+      "loss": 0.4914,
+      "step": 17783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.805776184271393e-06,
+      "loss": 0.4419,
+      "step": 17784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8047141993785905e-06,
+      "loss": 0.4549,
+      "step": 17785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803652254618751e-06,
+      "loss": 0.4063,
+      "step": 17786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.80259035000522e-06,
+      "loss": 0.4568,
+      "step": 17787
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801528485551339e-06,
+      "loss": 0.4956,
+      "step": 17788
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800466661270433e-06,
+      "loss": 0.4313,
+      "step": 17789
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799404877175847e-06,
+      "loss": 0.4838,
+      "step": 17790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798343133280918e-06,
+      "loss": 0.3842,
+      "step": 17791
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797281429598974e-06,
+      "loss": 0.426,
+      "step": 17792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7962197661433546e-06,
+      "loss": 0.4305,
+      "step": 17793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795158142927398e-06,
+      "loss": 0.4689,
+      "step": 17794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.794096559964434e-06,
+      "loss": 0.4859,
+      "step": 17795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793035017267793e-06,
+      "loss": 0.4105,
+      "step": 17796
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.791973514850817e-06,
+      "loss": 0.4424,
+      "step": 17797
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.790912052726829e-06,
+      "loss": 0.3953,
+      "step": 17798
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789850630909165e-06,
+      "loss": 0.4608,
+      "step": 17799
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78878924941116e-06,
+      "loss": 0.4426,
+      "step": 17800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787727908246142e-06,
+      "loss": 0.4447,
+      "step": 17801
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.786666607427439e-06,
+      "loss": 0.4589,
+      "step": 17802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.4591,
+      "step": 17803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.784544126882308e-06,
+      "loss": 0.4406,
+      "step": 17804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783482947182536e-06,
+      "loss": 0.4156,
+      "step": 17805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.782421807882399e-06,
+      "loss": 0.3957,
+      "step": 17806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.781360708995224e-06,
+      "loss": 0.4165,
+      "step": 17807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.780299650534337e-06,
+      "loss": 0.436,
+      "step": 17808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.77923863251307e-06,
+      "loss": 0.4277,
+      "step": 17809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7781776549447435e-06,
+      "loss": 0.4668,
+      "step": 17810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.777116717842686e-06,
+      "loss": 0.4649,
+      "step": 17811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776055821220228e-06,
+      "loss": 0.4642,
+      "step": 17812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.774994965090683e-06,
+      "loss": 0.4083,
+      "step": 17813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773934149467382e-06,
+      "loss": 0.3871,
+      "step": 17814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772873374363655e-06,
+      "loss": 0.4768,
+      "step": 17815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7718126397928134e-06,
+      "loss": 0.5277,
+      "step": 17816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770751945768188e-06,
+      "loss": 0.4732,
+      "step": 17817
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769691292303101e-06,
+      "loss": 0.463,
+      "step": 17818
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7686306794108685e-06,
+      "loss": 0.4455,
+      "step": 17819
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767570107104815e-06,
+      "loss": 0.4256,
+      "step": 17820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7665095753982665e-06,
+      "loss": 0.3936,
+      "step": 17821
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765449084304537e-06,
+      "loss": 0.4507,
+      "step": 17822
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.764388633836947e-06,
+      "loss": 0.4711,
+      "step": 17823
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763328224008815e-06,
+      "loss": 0.4624,
+      "step": 17824
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.762267854833466e-06,
+      "loss": 0.4725,
+      "step": 17825
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761207526324211e-06,
+      "loss": 0.4245,
+      "step": 17826
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760147238494374e-06,
+      "loss": 0.3948,
+      "step": 17827
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7590869913572685e-06,
+      "loss": 0.4615,
+      "step": 17828
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75802678492621e-06,
+      "loss": 0.4927,
+      "step": 17829
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756966619214515e-06,
+      "loss": 0.4298,
+      "step": 17830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7559064942355065e-06,
+      "loss": 0.4224,
+      "step": 17831
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.754846410002492e-06,
+      "loss": 0.4775,
+      "step": 17832
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.753786366528785e-06,
+      "loss": 0.4076,
+      "step": 17833
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7527263638277085e-06,
+      "loss": 0.4604,
+      "step": 17834
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751666401912566e-06,
+      "loss": 0.453,
+      "step": 17835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750606480796675e-06,
+      "loss": 0.4357,
+      "step": 17836
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749546600493353e-06,
+      "loss": 0.4092,
+      "step": 17837
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.748486761015907e-06,
+      "loss": 0.449,
+      "step": 17838
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7474269623776456e-06,
+      "loss": 0.3786,
+      "step": 17839
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746367204591888e-06,
+      "loss": 0.4475,
+      "step": 17840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745307487671937e-06,
+      "loss": 0.4398,
+      "step": 17841
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.744247811631109e-06,
+      "loss": 0.4166,
+      "step": 17842
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743188176482709e-06,
+      "loss": 0.4881,
+      "step": 17843
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.742128582240048e-06,
+      "loss": 0.4475,
+      "step": 17844
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741069028916434e-06,
+      "loss": 0.4491,
+      "step": 17845
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.740009516525179e-06,
+      "loss": 0.4409,
+      "step": 17846
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.738950045079582e-06,
+      "loss": 0.4328,
+      "step": 17847
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737890614592958e-06,
+      "loss": 0.4305,
+      "step": 17848
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.736831225078613e-06,
+      "loss": 0.4838,
+      "step": 17849
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7357718765498455e-06,
+      "loss": 0.4909,
+      "step": 17850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7347125690199665e-06,
+      "loss": 0.4871,
+      "step": 17851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733653302502285e-06,
+      "loss": 0.4498,
+      "step": 17852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732594077010096e-06,
+      "loss": 0.4389,
+      "step": 17853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7315348925567105e-06,
+      "loss": 0.41,
+      "step": 17854
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.730475749155433e-06,
+      "loss": 0.4325,
+      "step": 17855
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729416646819558e-06,
+      "loss": 0.4456,
+      "step": 17856
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.728357585562394e-06,
+      "loss": 0.4729,
+      "step": 17857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.727298565397244e-06,
+      "loss": 0.4794,
+      "step": 17858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726239586337408e-06,
+      "loss": 0.4092,
+      "step": 17859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.725180648396183e-06,
+      "loss": 0.4419,
+      "step": 17860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724121751586871e-06,
+      "loss": 0.4344,
+      "step": 17861
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.72306289592278e-06,
+      "loss": 0.42,
+      "step": 17862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.722004081417197e-06,
+      "loss": 0.4572,
+      "step": 17863
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7209453080834284e-06,
+      "loss": 0.4706,
+      "step": 17864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.719886575934773e-06,
+      "loss": 0.4172,
+      "step": 17865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718827884984522e-06,
+      "loss": 0.412,
+      "step": 17866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.717769235245976e-06,
+      "loss": 0.4487,
+      "step": 17867
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716710626732436e-06,
+      "loss": 0.4343,
+      "step": 17868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.715652059457192e-06,
+      "loss": 0.4501,
+      "step": 17869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.714593533433541e-06,
+      "loss": 0.4271,
+      "step": 17870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.713535048674783e-06,
+      "loss": 0.4133,
+      "step": 17871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.712476605194205e-06,
+      "loss": 0.4621,
+      "step": 17872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.711418203005104e-06,
+      "loss": 0.446,
+      "step": 17873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.710359842120779e-06,
+      "loss": 0.4468,
+      "step": 17874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.709301522554517e-06,
+      "loss": 0.438,
+      "step": 17875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.708243244319611e-06,
+      "loss": 0.4267,
+      "step": 17876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707185007429358e-06,
+      "loss": 0.4431,
+      "step": 17877
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70612681189704e-06,
+      "loss": 0.4358,
+      "step": 17878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705068657735957e-06,
+      "loss": 0.4818,
+      "step": 17879
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.704010544959396e-06,
+      "loss": 0.4686,
+      "step": 17880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.702952473580647e-06,
+      "loss": 0.4372,
+      "step": 17881
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701894443612997e-06,
+      "loss": 0.4531,
+      "step": 17882
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700836455069744e-06,
+      "loss": 0.4257,
+      "step": 17883
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699778507964163e-06,
+      "loss": 0.4551,
+      "step": 17884
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6987206023095515e-06,
+      "loss": 0.4593,
+      "step": 17885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6976627381191974e-06,
+      "loss": 0.4466,
+      "step": 17886
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.69660491540638e-06,
+      "loss": 0.4163,
+      "step": 17887
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.695547134184388e-06,
+      "loss": 0.4618,
+      "step": 17888
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694489394466515e-06,
+      "loss": 0.4287,
+      "step": 17889
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693431696266035e-06,
+      "loss": 0.4278,
+      "step": 17890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692374039596241e-06,
+      "loss": 0.4047,
+      "step": 17891
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.691316424470415e-06,
+      "loss": 0.4616,
+      "step": 17892
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690258850901838e-06,
+      "loss": 0.4087,
+      "step": 17893
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6892013189037945e-06,
+      "loss": 0.4648,
+      "step": 17894
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.688143828489569e-06,
+      "loss": 0.4539,
+      "step": 17895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687086379672445e-06,
+      "loss": 0.4404,
+      "step": 17896
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6860289724656975e-06,
+      "loss": 0.4111,
+      "step": 17897
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684971606882612e-06,
+      "loss": 0.4409,
+      "step": 17898
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6839142829364724e-06,
+      "loss": 0.4424,
+      "step": 17899
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6828570006405524e-06,
+      "loss": 0.4617,
+      "step": 17900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.681799760008134e-06,
+      "loss": 0.4955,
+      "step": 17901
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6807425610525e-06,
+      "loss": 0.4257,
+      "step": 17902
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.679685403786922e-06,
+      "loss": 0.4829,
+      "step": 17903
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.67862828822468e-06,
+      "loss": 0.4375,
+      "step": 17904
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677571214379058e-06,
+      "loss": 0.4357,
+      "step": 17905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676514182263324e-06,
+      "loss": 0.4341,
+      "step": 17906
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.675457191890756e-06,
+      "loss": 0.4604,
+      "step": 17907
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.674400243274637e-06,
+      "loss": 0.4436,
+      "step": 17908
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6733433364282325e-06,
+      "loss": 0.4364,
+      "step": 17909
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6722864713648215e-06,
+      "loss": 0.4731,
+      "step": 17910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671229648097685e-06,
+      "loss": 0.3833,
+      "step": 17911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6701728666400864e-06,
+      "loss": 0.447,
+      "step": 17912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6691161270053016e-06,
+      "loss": 0.4317,
+      "step": 17913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.668059429206611e-06,
+      "loss": 0.4032,
+      "step": 17914
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667002773257273e-06,
+      "loss": 0.4051,
+      "step": 17915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.665946159170571e-06,
+      "loss": 0.4341,
+      "step": 17916
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664889586959773e-06,
+      "loss": 0.4874,
+      "step": 17917
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6638330566381446e-06,
+      "loss": 0.4323,
+      "step": 17918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.66277656821896e-06,
+      "loss": 0.4347,
+      "step": 17919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6617201217154935e-06,
+      "loss": 0.4542,
+      "step": 17920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.660663717141005e-06,
+      "loss": 0.4073,
+      "step": 17921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.659607354508768e-06,
+      "loss": 0.4605,
+      "step": 17922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.658551033832052e-06,
+      "loss": 0.378,
+      "step": 17923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657494755124119e-06,
+      "loss": 0.459,
+      "step": 17924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.656438518398238e-06,
+      "loss": 0.4245,
+      "step": 17925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655382323667683e-06,
+      "loss": 0.4188,
+      "step": 17926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.654326170945709e-06,
+      "loss": 0.3743,
+      "step": 17927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653270060245585e-06,
+      "loss": 0.4286,
+      "step": 17928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6522139915805785e-06,
+      "loss": 0.4785,
+      "step": 17929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651157964963949e-06,
+      "loss": 0.4493,
+      "step": 17930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650101980408963e-06,
+      "loss": 0.4623,
+      "step": 17931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649046037928885e-06,
+      "loss": 0.4779,
+      "step": 17932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.647990137536979e-06,
+      "loss": 0.4274,
+      "step": 17933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.646934279246502e-06,
+      "loss": 0.4381,
+      "step": 17934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645878463070716e-06,
+      "loss": 0.4666,
+      "step": 17935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644822689022889e-06,
+      "loss": 0.422,
+      "step": 17936
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643766957116276e-06,
+      "loss": 0.4921,
+      "step": 17937
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.642711267364135e-06,
+      "loss": 0.4333,
+      "step": 17938
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.641655619779733e-06,
+      "loss": 0.4707,
+      "step": 17939
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6406000143763205e-06,
+      "loss": 0.4094,
+      "step": 17940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6395444511671595e-06,
+      "loss": 0.3878,
+      "step": 17941
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.638488930165513e-06,
+      "loss": 0.4404,
+      "step": 17942
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637433451384631e-06,
+      "loss": 0.4116,
+      "step": 17943
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.636378014837772e-06,
+      "loss": 0.4878,
+      "step": 17944
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.635322620538197e-06,
+      "loss": 0.4673,
+      "step": 17945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634267268499155e-06,
+      "loss": 0.3933,
+      "step": 17946
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633211958733905e-06,
+      "loss": 0.3743,
+      "step": 17947
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6321566912557035e-06,
+      "loss": 0.4353,
+      "step": 17948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631101466077801e-06,
+      "loss": 0.4254,
+      "step": 17949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630046283213452e-06,
+      "loss": 0.4361,
+      "step": 17950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6289911426759135e-06,
+      "loss": 0.4345,
+      "step": 17951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627936044478432e-06,
+      "loss": 0.4659,
+      "step": 17952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6268809886342634e-06,
+      "loss": 0.4589,
+      "step": 17953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625825975156661e-06,
+      "loss": 0.44,
+      "step": 17954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624771004058869e-06,
+      "loss": 0.3906,
+      "step": 17955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6237160753541425e-06,
+      "loss": 0.4251,
+      "step": 17956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622661189055735e-06,
+      "loss": 0.4386,
+      "step": 17957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621606345176888e-06,
+      "loss": 0.4722,
+      "step": 17958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.620551543730858e-06,
+      "loss": 0.4018,
+      "step": 17959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6194967847308896e-06,
+      "loss": 0.4441,
+      "step": 17960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.618442068190228e-06,
+      "loss": 0.4458,
+      "step": 17961
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.617387394122124e-06,
+      "loss": 0.4528,
+      "step": 17962
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616332762539827e-06,
+      "loss": 0.4652,
+      "step": 17963
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.615278173456578e-06,
+      "loss": 0.4324,
+      "step": 17964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614223626885623e-06,
+      "loss": 0.4446,
+      "step": 17965
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.613169122840213e-06,
+      "loss": 0.4534,
+      "step": 17966
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.612114661333584e-06,
+      "loss": 0.4413,
+      "step": 17967
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.611060242378986e-06,
+      "loss": 0.4657,
+      "step": 17968
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610005865989661e-06,
+      "loss": 0.4636,
+      "step": 17969
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608951532178856e-06,
+      "loss": 0.4499,
+      "step": 17970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.607897240959804e-06,
+      "loss": 0.4417,
+      "step": 17971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.606842992345755e-06,
+      "loss": 0.4232,
+      "step": 17972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60578878634995e-06,
+      "loss": 0.4591,
+      "step": 17973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.604734622985627e-06,
+      "loss": 0.4003,
+      "step": 17974
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.603680502266024e-06,
+      "loss": 0.4374,
+      "step": 17975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60262642420439e-06,
+      "loss": 0.4435,
+      "step": 17976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.601572388813953e-06,
+      "loss": 0.4031,
+      "step": 17977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600518396107958e-06,
+      "loss": 0.4406,
+      "step": 17978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599464446099646e-06,
+      "loss": 0.4326,
+      "step": 17979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598410538802248e-06,
+      "loss": 0.4489,
+      "step": 17980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597356674229003e-06,
+      "loss": 0.436,
+      "step": 17981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5963028523931525e-06,
+      "loss": 0.4682,
+      "step": 17982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595249073307926e-06,
+      "loss": 0.4121,
+      "step": 17983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594195336986562e-06,
+      "loss": 0.4635,
+      "step": 17984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593141643442296e-06,
+      "loss": 0.454,
+      "step": 17985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592087992688363e-06,
+      "loss": 0.4823,
+      "step": 17986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.591034384737992e-06,
+      "loss": 0.3997,
+      "step": 17987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5899808196044245e-06,
+      "loss": 0.4481,
+      "step": 17988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588927297300885e-06,
+      "loss": 0.4278,
+      "step": 17989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587873817840612e-06,
+      "loss": 0.4442,
+      "step": 17990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586820381236836e-06,
+      "loss": 0.4523,
+      "step": 17991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585766987502784e-06,
+      "loss": 0.4269,
+      "step": 17992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.584713636651689e-06,
+      "loss": 0.4189,
+      "step": 17993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.583660328696788e-06,
+      "loss": 0.4315,
+      "step": 17994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582607063651298e-06,
+      "loss": 0.4528,
+      "step": 17995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.581553841528458e-06,
+      "loss": 0.4212,
+      "step": 17996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580500662341496e-06,
+      "loss": 0.4512,
+      "step": 17997
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579447526103633e-06,
+      "loss": 0.3583,
+      "step": 17998
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578394432828099e-06,
+      "loss": 0.398,
+      "step": 17999
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5773413825281285e-06,
+      "loss": 0.3935,
+      "step": 18000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5762883752169394e-06,
+      "loss": 0.4744,
+      "step": 18001
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575235410907756e-06,
+      "loss": 0.5008,
+      "step": 18002
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574182489613814e-06,
+      "loss": 0.4689,
+      "step": 18003
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.573129611348328e-06,
+      "loss": 0.451,
+      "step": 18004
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5720767761245255e-06,
+      "loss": 0.3968,
+      "step": 18005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.571023983955631e-06,
+      "loss": 0.4413,
+      "step": 18006
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569971234854871e-06,
+      "loss": 0.4597,
+      "step": 18007
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568918528835459e-06,
+      "loss": 0.4688,
+      "step": 18008
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.567865865910624e-06,
+      "loss": 0.4122,
+      "step": 18009
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.566813246093591e-06,
+      "loss": 0.4301,
+      "step": 18010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.565760669397572e-06,
+      "loss": 0.3957,
+      "step": 18011
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.564708135835789e-06,
+      "loss": 0.4446,
+      "step": 18012
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5636556454214695e-06,
+      "loss": 0.4832,
+      "step": 18013
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562603198167821e-06,
+      "loss": 0.4423,
+      "step": 18014
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561550794088072e-06,
+      "loss": 0.4322,
+      "step": 18015
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56049843319544e-06,
+      "loss": 0.4187,
+      "step": 18016
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559446115503138e-06,
+      "loss": 0.394,
+      "step": 18017
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558393841024384e-06,
+      "loss": 0.4916,
+      "step": 18018
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5573416097724e-06,
+      "loss": 0.4231,
+      "step": 18019
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556289421760396e-06,
+      "loss": 0.4462,
+      "step": 18020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555237277001589e-06,
+      "loss": 0.4234,
+      "step": 18021
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.554185175509196e-06,
+      "loss": 0.433,
+      "step": 18022
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.553133117296431e-06,
+      "loss": 0.4494,
+      "step": 18023
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552081102376504e-06,
+      "loss": 0.4828,
+      "step": 18024
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.551029130762636e-06,
+      "loss": 0.4676,
+      "step": 18025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54997720246803e-06,
+      "loss": 0.4348,
+      "step": 18026
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548925317505908e-06,
+      "loss": 0.4884,
+      "step": 18027
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547873475889477e-06,
+      "loss": 0.3925,
+      "step": 18028
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5468216776319445e-06,
+      "loss": 0.4629,
+      "step": 18029
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545769922746526e-06,
+      "loss": 0.4504,
+      "step": 18030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544718211246434e-06,
+      "loss": 0.444,
+      "step": 18031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543666543144872e-06,
+      "loss": 0.4164,
+      "step": 18032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.542614918455054e-06,
+      "loss": 0.4481,
+      "step": 18033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5415633371901865e-06,
+      "loss": 0.3979,
+      "step": 18034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540511799363473e-06,
+      "loss": 0.424,
+      "step": 18035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539460304988128e-06,
+      "loss": 0.4067,
+      "step": 18036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.538408854077357e-06,
+      "loss": 0.4505,
+      "step": 18037
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537357446644364e-06,
+      "loss": 0.424,
+      "step": 18038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536306082702353e-06,
+      "loss": 0.4497,
+      "step": 18039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.535254762264533e-06,
+      "loss": 0.4576,
+      "step": 18040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.534203485344106e-06,
+      "loss": 0.4397,
+      "step": 18041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5331522519542776e-06,
+      "loss": 0.4452,
+      "step": 18042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.532101062108251e-06,
+      "loss": 0.4353,
+      "step": 18043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531049915819234e-06,
+      "loss": 0.4402,
+      "step": 18044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5299988131004174e-06,
+      "loss": 0.4388,
+      "step": 18045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.528947753965013e-06,
+      "loss": 0.4388,
+      "step": 18046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5278967384262215e-06,
+      "loss": 0.4989,
+      "step": 18047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52684576649724e-06,
+      "loss": 0.4319,
+      "step": 18048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.525794838191267e-06,
+      "loss": 0.4733,
+      "step": 18049
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524743953521511e-06,
+      "loss": 0.4422,
+      "step": 18050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52369311250116e-06,
+      "loss": 0.4357,
+      "step": 18051
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522642315143419e-06,
+      "loss": 0.4352,
+      "step": 18052
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52159156146149e-06,
+      "loss": 0.4245,
+      "step": 18053
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520540851468563e-06,
+      "loss": 0.4818,
+      "step": 18054
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.519490185177836e-06,
+      "loss": 0.4174,
+      "step": 18055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518439562602511e-06,
+      "loss": 0.4347,
+      "step": 18056
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517388983755777e-06,
+      "loss": 0.4211,
+      "step": 18057
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516338448650832e-06,
+      "loss": 0.4294,
+      "step": 18058
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515287957300876e-06,
+      "loss": 0.4317,
+      "step": 18059
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.514237509719093e-06,
+      "loss": 0.382,
+      "step": 18060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513187105918682e-06,
+      "loss": 0.4791,
+      "step": 18061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.51213674591284e-06,
+      "loss": 0.4766,
+      "step": 18062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.511086429714753e-06,
+      "loss": 0.4514,
+      "step": 18063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510036157337617e-06,
+      "loss": 0.4547,
+      "step": 18064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.508985928794624e-06,
+      "loss": 0.4254,
+      "step": 18065
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507935744098961e-06,
+      "loss": 0.3793,
+      "step": 18066
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.506885603263819e-06,
+      "loss": 0.3791,
+      "step": 18067
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.505835506302394e-06,
+      "loss": 0.4353,
+      "step": 18068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504785453227868e-06,
+      "loss": 0.4199,
+      "step": 18069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503735444053432e-06,
+      "loss": 0.4888,
+      "step": 18070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.502685478792277e-06,
+      "loss": 0.475,
+      "step": 18071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501635557457585e-06,
+      "loss": 0.4178,
+      "step": 18072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5005856800625456e-06,
+      "loss": 0.4386,
+      "step": 18073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499535846620352e-06,
+      "loss": 0.4473,
+      "step": 18074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49848605714418e-06,
+      "loss": 0.4467,
+      "step": 18075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497436311647219e-06,
+      "loss": 0.4574,
+      "step": 18076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496386610142653e-06,
+      "loss": 0.4569,
+      "step": 18077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495336952643672e-06,
+      "loss": 0.4729,
+      "step": 18078
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494287339163452e-06,
+      "loss": 0.5136,
+      "step": 18079
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493237769715178e-06,
+      "loss": 0.4231,
+      "step": 18080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.492188244312039e-06,
+      "loss": 0.4643,
+      "step": 18081
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491138762967206e-06,
+      "loss": 0.4425,
+      "step": 18082
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490089325693869e-06,
+      "loss": 0.4189,
+      "step": 18083
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48903993250521e-06,
+      "loss": 0.4164,
+      "step": 18084
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487990583414403e-06,
+      "loss": 0.4419,
+      "step": 18085
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48694127843463e-06,
+      "loss": 0.4199,
+      "step": 18086
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485892017579075e-06,
+      "loss": 0.4718,
+      "step": 18087
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484842800860909e-06,
+      "loss": 0.4324,
+      "step": 18088
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4837936282933155e-06,
+      "loss": 0.407,
+      "step": 18089
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.482744499889473e-06,
+      "loss": 0.3842,
+      "step": 18090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.481695415662554e-06,
+      "loss": 0.4181,
+      "step": 18091
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4806463756257376e-06,
+      "loss": 0.4976,
+      "step": 18092
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479597379792204e-06,
+      "loss": 0.4732,
+      "step": 18093
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.47854842817512e-06,
+      "loss": 0.5124,
+      "step": 18094
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4774995207876654e-06,
+      "loss": 0.4607,
+      "step": 18095
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476450657643019e-06,
+      "loss": 0.4418,
+      "step": 18096
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.475401838754343e-06,
+      "loss": 0.5168,
+      "step": 18097
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474353064134817e-06,
+      "loss": 0.419,
+      "step": 18098
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.473304333797621e-06,
+      "loss": 0.456,
+      "step": 18099
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.472255647755914e-06,
+      "loss": 0.43,
+      "step": 18100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.471207006022876e-06,
+      "loss": 0.4708,
+      "step": 18101
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4701584086116774e-06,
+      "loss": 0.4334,
+      "step": 18102
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469109855535483e-06,
+      "loss": 0.4076,
+      "step": 18103
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468061346807467e-06,
+      "loss": 0.4216,
+      "step": 18104
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.467012882440802e-06,
+      "loss": 0.4377,
+      "step": 18105
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465964462448651e-06,
+      "loss": 0.4928,
+      "step": 18106
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464916086844182e-06,
+      "loss": 0.4759,
+      "step": 18107
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.46386775564057e-06,
+      "loss": 0.4085,
+      "step": 18108
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462819468850972e-06,
+      "loss": 0.4508,
+      "step": 18109
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.46177122648856e-06,
+      "loss": 0.4694,
+      "step": 18110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460723028566503e-06,
+      "loss": 0.4194,
+      "step": 18111
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459674875097962e-06,
+      "loss": 0.4023,
+      "step": 18112
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4586267660961e-06,
+      "loss": 0.4697,
+      "step": 18113
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4575787015740846e-06,
+      "loss": 0.484,
+      "step": 18114
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456530681545083e-06,
+      "loss": 0.4825,
+      "step": 18115
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.455482706022252e-06,
+      "loss": 0.38,
+      "step": 18116
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454434775018757e-06,
+      "loss": 0.4135,
+      "step": 18117
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453386888547762e-06,
+      "loss": 0.4512,
+      "step": 18118
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45233904662242e-06,
+      "loss": 0.4889,
+      "step": 18119
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4512912492559025e-06,
+      "loss": 0.4053,
+      "step": 18120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450243496461366e-06,
+      "loss": 0.4424,
+      "step": 18121
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.44919578825197e-06,
+      "loss": 0.3879,
+      "step": 18122
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448148124640872e-06,
+      "loss": 0.4433,
+      "step": 18123
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.447100505641235e-06,
+      "loss": 0.4915,
+      "step": 18124
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446052931266213e-06,
+      "loss": 0.4186,
+      "step": 18125
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445005401528965e-06,
+      "loss": 0.4493,
+      "step": 18126
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.443957916442649e-06,
+      "loss": 0.4288,
+      "step": 18127
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442910476020421e-06,
+      "loss": 0.4343,
+      "step": 18128
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.441863080275435e-06,
+      "loss": 0.47,
+      "step": 18129
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.44081572922085e-06,
+      "loss": 0.4686,
+      "step": 18130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439768422869816e-06,
+      "loss": 0.4685,
+      "step": 18131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438721161235493e-06,
+      "loss": 0.3849,
+      "step": 18132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437673944331033e-06,
+      "loss": 0.4258,
+      "step": 18133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436626772169584e-06,
+      "loss": 0.4547,
+      "step": 18134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4355796447643025e-06,
+      "loss": 0.4469,
+      "step": 18135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434532562128344e-06,
+      "loss": 0.4657,
+      "step": 18136
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.433485524274852e-06,
+      "loss": 0.4728,
+      "step": 18137
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4324385312169825e-06,
+      "loss": 0.4709,
+      "step": 18138
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431391582967888e-06,
+      "loss": 0.4184,
+      "step": 18139
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430344679540711e-06,
+      "loss": 0.4205,
+      "step": 18140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429297820948605e-06,
+      "loss": 0.4158,
+      "step": 18141
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4282510072047225e-06,
+      "loss": 0.4043,
+      "step": 18142
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427204238322204e-06,
+      "loss": 0.4193,
+      "step": 18143
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4261575143141995e-06,
+      "loss": 0.4516,
+      "step": 18144
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4251108351938606e-06,
+      "loss": 0.4257,
+      "step": 18145
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.424064200974326e-06,
+      "loss": 0.5115,
+      "step": 18146
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.423017611668745e-06,
+      "loss": 0.4391,
+      "step": 18147
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.421971067290267e-06,
+      "loss": 0.4463,
+      "step": 18148
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420924567852032e-06,
+      "loss": 0.4215,
+      "step": 18149
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419878113367182e-06,
+      "loss": 0.4545,
+      "step": 18150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418831703848862e-06,
+      "loss": 0.4373,
+      "step": 18151
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.417785339310223e-06,
+      "loss": 0.4774,
+      "step": 18152
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.416739019764397e-06,
+      "loss": 0.3736,
+      "step": 18153
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.415692745224528e-06,
+      "loss": 0.4048,
+      "step": 18154
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414646515703763e-06,
+      "loss": 0.4695,
+      "step": 18155
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413600331215234e-06,
+      "loss": 0.4404,
+      "step": 18156
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.412554191772087e-06,
+      "loss": 0.4621,
+      "step": 18157
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.411508097387462e-06,
+      "loss": 0.4651,
+      "step": 18158
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.410462048074498e-06,
+      "loss": 0.4572,
+      "step": 18159
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409416043846326e-06,
+      "loss": 0.4188,
+      "step": 18160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.408370084716096e-06,
+      "loss": 0.4537,
+      "step": 18161
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4073241706969335e-06,
+      "loss": 0.4514,
+      "step": 18162
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.406278301801983e-06,
+      "loss": 0.478,
+      "step": 18163
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405232478044379e-06,
+      "loss": 0.4893,
+      "step": 18164
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.404186699437255e-06,
+      "loss": 0.4768,
+      "step": 18165
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403140965993746e-06,
+      "loss": 0.3836,
+      "step": 18166
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.402095277726992e-06,
+      "loss": 0.4356,
+      "step": 18167
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401049634650119e-06,
+      "loss": 0.437,
+      "step": 18168
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400004036776266e-06,
+      "loss": 0.4689,
+      "step": 18169
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.398958484118566e-06,
+      "loss": 0.441,
+      "step": 18170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397912976690146e-06,
+      "loss": 0.4656,
+      "step": 18171
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.396867514504139e-06,
+      "loss": 0.4015,
+      "step": 18172
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3958220975736825e-06,
+      "loss": 0.4268,
+      "step": 18173
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.394776725911899e-06,
+      "loss": 0.4239,
+      "step": 18174
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.393731399531923e-06,
+      "loss": 0.4127,
+      "step": 18175
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392686118446885e-06,
+      "loss": 0.4622,
+      "step": 18176
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.391640882669907e-06,
+      "loss": 0.4557,
+      "step": 18177
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390595692214122e-06,
+      "loss": 0.4228,
+      "step": 18178
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389550547092661e-06,
+      "loss": 0.4634,
+      "step": 18179
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.388505447318644e-06,
+      "loss": 0.4227,
+      "step": 18180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387460392905202e-06,
+      "loss": 0.4504,
+      "step": 18181
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386415383865462e-06,
+      "loss": 0.4189,
+      "step": 18182
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.385370420212542e-06,
+      "loss": 0.4257,
+      "step": 18183
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384325501959574e-06,
+      "loss": 0.4363,
+      "step": 18184
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3832806291196835e-06,
+      "loss": 0.4237,
+      "step": 18185
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3822358017059895e-06,
+      "loss": 0.453,
+      "step": 18186
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3811910197316126e-06,
+      "loss": 0.4295,
+      "step": 18187
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.380146283209681e-06,
+      "loss": 0.412,
+      "step": 18188
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379101592153318e-06,
+      "loss": 0.4501,
+      "step": 18189
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37805694657564e-06,
+      "loss": 0.4932,
+      "step": 18190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377012346489767e-06,
+      "loss": 0.3952,
+      "step": 18191
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375967791908826e-06,
+      "loss": 0.4204,
+      "step": 18192
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3749232828459285e-06,
+      "loss": 0.3977,
+      "step": 18193
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3738788193142005e-06,
+      "loss": 0.4377,
+      "step": 18194
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372834401326757e-06,
+      "loss": 0.4336,
+      "step": 18195
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.371790028896716e-06,
+      "loss": 0.4704,
+      "step": 18196
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370745702037195e-06,
+      "loss": 0.3941,
+      "step": 18197
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369701420761314e-06,
+      "loss": 0.4664,
+      "step": 18198
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368657185082183e-06,
+      "loss": 0.4956,
+      "step": 18199
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.367612995012922e-06,
+      "loss": 0.4253,
+      "step": 18200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366568850566648e-06,
+      "loss": 0.4736,
+      "step": 18201
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365524751756469e-06,
+      "loss": 0.4121,
+      "step": 18202
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364480698595503e-06,
+      "loss": 0.4565,
+      "step": 18203
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.363436691096867e-06,
+      "loss": 0.3946,
+      "step": 18204
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362392729273665e-06,
+      "loss": 0.4339,
+      "step": 18205
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361348813139016e-06,
+      "loss": 0.4187,
+      "step": 18206
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.360304942706033e-06,
+      "loss": 0.437,
+      "step": 18207
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359261117987818e-06,
+      "loss": 0.478,
+      "step": 18208
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3582173389974876e-06,
+      "loss": 0.4901,
+      "step": 18209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.357173605748157e-06,
+      "loss": 0.3985,
+      "step": 18210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356129918252927e-06,
+      "loss": 0.4053,
+      "step": 18211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3550862765249064e-06,
+      "loss": 0.472,
+      "step": 18212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.354042680577212e-06,
+      "loss": 0.3778,
+      "step": 18213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35299913042294e-06,
+      "loss": 0.4584,
+      "step": 18214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351955626075204e-06,
+      "loss": 0.4488,
+      "step": 18215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350912167547114e-06,
+      "loss": 0.4749,
+      "step": 18216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.349868754851769e-06,
+      "loss": 0.4257,
+      "step": 18217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348825388002275e-06,
+      "loss": 0.4554,
+      "step": 18218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.347782067011743e-06,
+      "loss": 0.4017,
+      "step": 18219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346738791893268e-06,
+      "loss": 0.4515,
+      "step": 18220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345695562659961e-06,
+      "loss": 0.4745,
+      "step": 18221
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344652379324922e-06,
+      "loss": 0.4322,
+      "step": 18222
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.343609241901254e-06,
+      "loss": 0.4238,
+      "step": 18223
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342566150402056e-06,
+      "loss": 0.4311,
+      "step": 18224
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.341523104840434e-06,
+      "loss": 0.444,
+      "step": 18225
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34048010522949e-06,
+      "loss": 0.4369,
+      "step": 18226
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339437151582317e-06,
+      "loss": 0.4218,
+      "step": 18227
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338394243912018e-06,
+      "loss": 0.4457,
+      "step": 18228
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.337351382231695e-06,
+      "loss": 0.4324,
+      "step": 18229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336308566554441e-06,
+      "loss": 0.4257,
+      "step": 18230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33526579689336e-06,
+      "loss": 0.4433,
+      "step": 18231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.334223073261543e-06,
+      "loss": 0.4641,
+      "step": 18232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333180395672092e-06,
+      "loss": 0.4315,
+      "step": 18233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332137764138097e-06,
+      "loss": 0.4024,
+      "step": 18234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331095178672661e-06,
+      "loss": 0.494,
+      "step": 18235
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.330052639288871e-06,
+      "loss": 0.498,
+      "step": 18236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329010145999827e-06,
+      "loss": 0.4864,
+      "step": 18237
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.327967698818625e-06,
+      "loss": 0.3993,
+      "step": 18238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3269252977583496e-06,
+      "loss": 0.4431,
+      "step": 18239
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.325882942832098e-06,
+      "loss": 0.4219,
+      "step": 18240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3248406340529665e-06,
+      "loss": 0.4607,
+      "step": 18241
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.323798371434038e-06,
+      "loss": 0.4488,
+      "step": 18242
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.32275615498841e-06,
+      "loss": 0.4544,
+      "step": 18243
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321713984729173e-06,
+      "loss": 0.4214,
+      "step": 18244
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3206718606694104e-06,
+      "loss": 0.3696,
+      "step": 18245
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319629782822215e-06,
+      "loss": 0.4402,
+      "step": 18246
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31858775120068e-06,
+      "loss": 0.4557,
+      "step": 18247
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317545765817887e-06,
+      "loss": 0.4711,
+      "step": 18248
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316503826686923e-06,
+      "loss": 0.4698,
+      "step": 18249
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3154619338208815e-06,
+      "loss": 0.4707,
+      "step": 18250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31442008723284e-06,
+      "loss": 0.4671,
+      "step": 18251
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31337828693589e-06,
+      "loss": 0.4958,
+      "step": 18252
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3123365329431175e-06,
+      "loss": 0.4749,
+      "step": 18253
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.311294825267603e-06,
+      "loss": 0.4428,
+      "step": 18254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.310253163922431e-06,
+      "loss": 0.4647,
+      "step": 18255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.309211548920688e-06,
+      "loss": 0.4563,
+      "step": 18256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.308169980275453e-06,
+      "loss": 0.4357,
+      "step": 18257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307128457999812e-06,
+      "loss": 0.4747,
+      "step": 18258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306086982106842e-06,
+      "loss": 0.4428,
+      "step": 18259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.305045552609628e-06,
+      "loss": 0.4119,
+      "step": 18260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304004169521246e-06,
+      "loss": 0.4079,
+      "step": 18261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30296283285478e-06,
+      "loss": 0.4535,
+      "step": 18262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3019215426233106e-06,
+      "loss": 0.4188,
+      "step": 18263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.300880298839912e-06,
+      "loss": 0.3784,
+      "step": 18264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.299839101517662e-06,
+      "loss": 0.5043,
+      "step": 18265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.298797950669644e-06,
+      "loss": 0.4582,
+      "step": 18266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.297756846308927e-06,
+      "loss": 0.4201,
+      "step": 18267
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296715788448594e-06,
+      "loss": 0.4482,
+      "step": 18268
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295674777101717e-06,
+      "loss": 0.4636,
+      "step": 18269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.294633812281374e-06,
+      "loss": 0.4377,
+      "step": 18270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2935928940006345e-06,
+      "loss": 0.4364,
+      "step": 18271
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29255202227258e-06,
+      "loss": 0.3723,
+      "step": 18272
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291511197110276e-06,
+      "loss": 0.4145,
+      "step": 18273
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2904704185268005e-06,
+      "loss": 0.4142,
+      "step": 18274
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289429686535226e-06,
+      "loss": 0.4685,
+      "step": 18275
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2883890011486204e-06,
+      "loss": 0.4638,
+      "step": 18276
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287348362380055e-06,
+      "loss": 0.4381,
+      "step": 18277
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2863077702426075e-06,
+      "loss": 0.462,
+      "step": 18278
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285267224749337e-06,
+      "loss": 0.518,
+      "step": 18279
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28422672591332e-06,
+      "loss": 0.4637,
+      "step": 18280
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283186273747628e-06,
+      "loss": 0.449,
+      "step": 18281
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.282145868265318e-06,
+      "loss": 0.4423,
+      "step": 18282
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281105509479465e-06,
+      "loss": 0.4784,
+      "step": 18283
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280065197403139e-06,
+      "loss": 0.4545,
+      "step": 18284
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2790249320494e-06,
+      "loss": 0.4341,
+      "step": 18285
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277984713431316e-06,
+      "loss": 0.4486,
+      "step": 18286
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276944541561954e-06,
+      "loss": 0.4483,
+      "step": 18287
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275904416454375e-06,
+      "loss": 0.4186,
+      "step": 18288
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274864338121645e-06,
+      "loss": 0.4586,
+      "step": 18289
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.27382430657683e-06,
+      "loss": 0.469,
+      "step": 18290
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.27278432183299e-06,
+      "loss": 0.4618,
+      "step": 18291
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271744383903185e-06,
+      "loss": 0.4558,
+      "step": 18292
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270704492800483e-06,
+      "loss": 0.4365,
+      "step": 18293
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.269664648537938e-06,
+      "loss": 0.4443,
+      "step": 18294
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268624851128615e-06,
+      "loss": 0.4516,
+      "step": 18295
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267585100585571e-06,
+      "loss": 0.4214,
+      "step": 18296
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2665453969218725e-06,
+      "loss": 0.4408,
+      "step": 18297
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.265505740150568e-06,
+      "loss": 0.4629,
+      "step": 18298
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264466130284721e-06,
+      "loss": 0.3867,
+      "step": 18299
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.263426567337391e-06,
+      "loss": 0.4125,
+      "step": 18300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262387051321632e-06,
+      "loss": 0.4131,
+      "step": 18301
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.261347582250499e-06,
+      "loss": 0.4681,
+      "step": 18302
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260308160137054e-06,
+      "loss": 0.4833,
+      "step": 18303
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259268784994344e-06,
+      "loss": 0.3958,
+      "step": 18304
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258229456835428e-06,
+      "loss": 0.4508,
+      "step": 18305
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.257190175673362e-06,
+      "loss": 0.46,
+      "step": 18306
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.256150941521195e-06,
+      "loss": 0.429,
+      "step": 18307
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.25511175439198e-06,
+      "loss": 0.4397,
+      "step": 18308
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254072614298776e-06,
+      "loss": 0.4799,
+      "step": 18309
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253033521254626e-06,
+      "loss": 0.4526,
+      "step": 18310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251994475272586e-06,
+      "loss": 0.4526,
+      "step": 18311
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250955476365708e-06,
+      "loss": 0.4487,
+      "step": 18312
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249916524547035e-06,
+      "loss": 0.407,
+      "step": 18313
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2488776198296195e-06,
+      "loss": 0.4136,
+      "step": 18314
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.247838762226516e-06,
+      "loss": 0.4155,
+      "step": 18315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246799951750764e-06,
+      "loss": 0.4507,
+      "step": 18316
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245761188415416e-06,
+      "loss": 0.4027,
+      "step": 18317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.24472247223352e-06,
+      "loss": 0.4692,
+      "step": 18318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.243683803218117e-06,
+      "loss": 0.4224,
+      "step": 18319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242645181382255e-06,
+      "loss": 0.4815,
+      "step": 18320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241606606738984e-06,
+      "loss": 0.4321,
+      "step": 18321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240568079301342e-06,
+      "loss": 0.4493,
+      "step": 18322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2395295990823744e-06,
+      "loss": 0.4955,
+      "step": 18323
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2384911660951294e-06,
+      "loss": 0.3898,
+      "step": 18324
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237452780352642e-06,
+      "loss": 0.4609,
+      "step": 18325
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236414441867959e-06,
+      "loss": 0.4574,
+      "step": 18326
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.235376150654124e-06,
+      "loss": 0.4052,
+      "step": 18327
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234337906724175e-06,
+      "loss": 0.4403,
+      "step": 18328
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233299710091149e-06,
+      "loss": 0.3816,
+      "step": 18329
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2322615607680945e-06,
+      "loss": 0.4539,
+      "step": 18330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231223458768042e-06,
+      "loss": 0.4365,
+      "step": 18331
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.230185404104035e-06,
+      "loss": 0.4803,
+      "step": 18332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22914739678911e-06,
+      "loss": 0.4595,
+      "step": 18333
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228109436836308e-06,
+      "loss": 0.423,
+      "step": 18334
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227071524258659e-06,
+      "loss": 0.4694,
+      "step": 18335
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226033659069204e-06,
+      "loss": 0.4339,
+      "step": 18336
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.224995841280979e-06,
+      "loss": 0.4755,
+      "step": 18337
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223958070907017e-06,
+      "loss": 0.4116,
+      "step": 18338
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22292034796035e-06,
+      "loss": 0.4082,
+      "step": 18339
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.221882672454022e-06,
+      "loss": 0.4299,
+      "step": 18340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2208450444010525e-06,
+      "loss": 0.4091,
+      "step": 18341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.219807463814485e-06,
+      "loss": 0.4433,
+      "step": 18342
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218769930707347e-06,
+      "loss": 0.4251,
+      "step": 18343
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2177324450926676e-06,
+      "loss": 0.4506,
+      "step": 18344
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.216695006983481e-06,
+      "loss": 0.4147,
+      "step": 18345
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.215657616392821e-06,
+      "loss": 0.4189,
+      "step": 18346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.214620273333709e-06,
+      "loss": 0.4155,
+      "step": 18347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213582977819179e-06,
+      "loss": 0.4841,
+      "step": 18348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212545729862263e-06,
+      "loss": 0.4226,
+      "step": 18349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2115085294759805e-06,
+      "loss": 0.4744,
+      "step": 18350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.210471376673361e-06,
+      "loss": 0.4661,
+      "step": 18351
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209434271467439e-06,
+      "loss": 0.4344,
+      "step": 18352
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.208397213871233e-06,
+      "loss": 0.4257,
+      "step": 18353
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2073602038977675e-06,
+      "loss": 0.4487,
+      "step": 18354
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206323241560075e-06,
+      "loss": 0.4223,
+      "step": 18355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20528632687117e-06,
+      "loss": 0.4924,
+      "step": 18356
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2042494598440825e-06,
+      "loss": 0.4563,
+      "step": 18357
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2032126404918384e-06,
+      "loss": 0.4626,
+      "step": 18358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.202175868827453e-06,
+      "loss": 0.4126,
+      "step": 18359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20113914486395e-06,
+      "loss": 0.4191,
+      "step": 18360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200102468614355e-06,
+      "loss": 0.4556,
+      "step": 18361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199065840091684e-06,
+      "loss": 0.4197,
+      "step": 18362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19802925930896e-06,
+      "loss": 0.4439,
+      "step": 18363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196992726279201e-06,
+      "loss": 0.452,
+      "step": 18364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.195956241015426e-06,
+      "loss": 0.4114,
+      "step": 18365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194919803530652e-06,
+      "loss": 0.4735,
+      "step": 18366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.193883413837904e-06,
+      "loss": 0.4592,
+      "step": 18367
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1928470719501876e-06,
+      "loss": 0.4628,
+      "step": 18368
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191810777880527e-06,
+      "loss": 0.43,
+      "step": 18369
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190774531641935e-06,
+      "loss": 0.4181,
+      "step": 18370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189738333247433e-06,
+      "loss": 0.4618,
+      "step": 18371
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.188702182710026e-06,
+      "loss": 0.4733,
+      "step": 18372
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187666080042736e-06,
+      "loss": 0.4488,
+      "step": 18373
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186630025258572e-06,
+      "loss": 0.4239,
+      "step": 18374
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185594018370549e-06,
+      "loss": 0.4839,
+      "step": 18375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184558059391676e-06,
+      "loss": 0.3632,
+      "step": 18376
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183522148334972e-06,
+      "loss": 0.4537,
+      "step": 18377
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.182486285213442e-06,
+      "loss": 0.4125,
+      "step": 18378
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181450470040098e-06,
+      "loss": 0.4516,
+      "step": 18379
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.180414702827951e-06,
+      "loss": 0.5466,
+      "step": 18380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179378983590008e-06,
+      "loss": 0.477,
+      "step": 18381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178343312339277e-06,
+      "loss": 0.4254,
+      "step": 18382
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177307689088772e-06,
+      "loss": 0.4478,
+      "step": 18383
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176272113851491e-06,
+      "loss": 0.3646,
+      "step": 18384
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.175236586640452e-06,
+      "loss": 0.4343,
+      "step": 18385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174201107468656e-06,
+      "loss": 0.4629,
+      "step": 18386
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.4338,
+      "step": 18387
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172130293294805e-06,
+      "loss": 0.4363,
+      "step": 18388
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171094958318769e-06,
+      "loss": 0.5003,
+      "step": 18389
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170059671433992e-06,
+      "loss": 0.4873,
+      "step": 18390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1690244326534786e-06,
+      "loss": 0.4729,
+      "step": 18391
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167989241990237e-06,
+      "loss": 0.4492,
+      "step": 18392
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166954099457261e-06,
+      "loss": 0.3799,
+      "step": 18393
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165919005067556e-06,
+      "loss": 0.4268,
+      "step": 18394
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164883958834127e-06,
+      "loss": 0.4586,
+      "step": 18395
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.163848960769967e-06,
+      "loss": 0.4763,
+      "step": 18396
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162814010888079e-06,
+      "loss": 0.4156,
+      "step": 18397
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161779109201465e-06,
+      "loss": 0.4444,
+      "step": 18398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1607442557231165e-06,
+      "loss": 0.4301,
+      "step": 18399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159709450466036e-06,
+      "loss": 0.4429,
+      "step": 18400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1586746934432214e-06,
+      "loss": 0.4154,
+      "step": 18401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157639984667667e-06,
+      "loss": 0.4093,
+      "step": 18402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156605324152369e-06,
+      "loss": 0.445,
+      "step": 18403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155570711910326e-06,
+      "loss": 0.4547,
+      "step": 18404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1545361479545266e-06,
+      "loss": 0.4524,
+      "step": 18405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153501632297971e-06,
+      "loss": 0.4348,
+      "step": 18406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152467164953647e-06,
+      "loss": 0.4152,
+      "step": 18407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151432745934556e-06,
+      "loss": 0.4368,
+      "step": 18408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1503983752536814e-06,
+      "loss": 0.4122,
+      "step": 18409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149364052924021e-06,
+      "loss": 0.4717,
+      "step": 18410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.148329778958561e-06,
+      "loss": 0.4183,
+      "step": 18411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147295553370298e-06,
+      "loss": 0.4856,
+      "step": 18412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.146261376172216e-06,
+      "loss": 0.449,
+      "step": 18413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14522724737731e-06,
+      "loss": 0.4291,
+      "step": 18414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1441931669985624e-06,
+      "loss": 0.4312,
+      "step": 18415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143159135048966e-06,
+      "loss": 0.3826,
+      "step": 18416
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.142125151541509e-06,
+      "loss": 0.4671,
+      "step": 18417
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141091216489172e-06,
+      "loss": 0.3956,
+      "step": 18418
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.140057329904945e-06,
+      "loss": 0.4087,
+      "step": 18419
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.139023491801819e-06,
+      "loss": 0.4496,
+      "step": 18420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13798970219277e-06,
+      "loss": 0.4881,
+      "step": 18421
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136955961090789e-06,
+      "loss": 0.4699,
+      "step": 18422
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.135922268508859e-06,
+      "loss": 0.4975,
+      "step": 18423
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.134888624459958e-06,
+      "loss": 0.4297,
+      "step": 18424
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.133855028957072e-06,
+      "loss": 0.3906,
+      "step": 18425
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132821482013187e-06,
+      "loss": 0.488,
+      "step": 18426
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13178798364128e-06,
+      "loss": 0.4535,
+      "step": 18427
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130754533854331e-06,
+      "loss": 0.4387,
+      "step": 18428
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129721132665324e-06,
+      "loss": 0.4516,
+      "step": 18429
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.128687780087232e-06,
+      "loss": 0.4049,
+      "step": 18430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12765447613304e-06,
+      "loss": 0.4566,
+      "step": 18431
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126621220815727e-06,
+      "loss": 0.4316,
+      "step": 18432
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.125588014148268e-06,
+      "loss": 0.4424,
+      "step": 18433
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1245548561436385e-06,
+      "loss": 0.4155,
+      "step": 18434
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.12352174681482e-06,
+      "loss": 0.4677,
+      "step": 18435
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.122488686174781e-06,
+      "loss": 0.4265,
+      "step": 18436
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121455674236505e-06,
+      "loss": 0.5046,
+      "step": 18437
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.120422711012962e-06,
+      "loss": 0.3977,
+      "step": 18438
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.119389796517128e-06,
+      "loss": 0.4521,
+      "step": 18439
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.118356930761971e-06,
+      "loss": 0.4797,
+      "step": 18440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117324113760474e-06,
+      "loss": 0.4329,
+      "step": 18441
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1162913455256005e-06,
+      "loss": 0.4782,
+      "step": 18442
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.115258626070326e-06,
+      "loss": 0.3995,
+      "step": 18443
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.11422595540762e-06,
+      "loss": 0.5234,
+      "step": 18444
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113193333550456e-06,
+      "loss": 0.4284,
+      "step": 18445
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.112160760511798e-06,
+      "loss": 0.4102,
+      "step": 18446
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1111282363046195e-06,
+      "loss": 0.4841,
+      "step": 18447
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.11009576094189e-06,
+      "loss": 0.5025,
+      "step": 18448
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.109063334436576e-06,
+      "loss": 0.4734,
+      "step": 18449
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10803095680164e-06,
+      "loss": 0.457,
+      "step": 18450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1069986280500595e-06,
+      "loss": 0.4151,
+      "step": 18451
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10596634819479e-06,
+      "loss": 0.4648,
+      "step": 18452
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104934117248802e-06,
+      "loss": 0.4563,
+      "step": 18453
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103901935225062e-06,
+      "loss": 0.4353,
+      "step": 18454
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.102869802136528e-06,
+      "loss": 0.4652,
+      "step": 18455
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101837717996168e-06,
+      "loss": 0.4403,
+      "step": 18456
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.100805682816948e-06,
+      "loss": 0.4665,
+      "step": 18457
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0997736966118245e-06,
+      "loss": 0.4301,
+      "step": 18458
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09874175939376e-06,
+      "loss": 0.4714,
+      "step": 18459
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097709871175723e-06,
+      "loss": 0.4218,
+      "step": 18460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.096678031970663e-06,
+      "loss": 0.4392,
+      "step": 18461
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095646241791546e-06,
+      "loss": 0.4125,
+      "step": 18462
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094614500651335e-06,
+      "loss": 0.3997,
+      "step": 18463
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.093582808562984e-06,
+      "loss": 0.423,
+      "step": 18464
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092551165539446e-06,
+      "loss": 0.443,
+      "step": 18465
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091519571593691e-06,
+      "loss": 0.4761,
+      "step": 18466
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090488026738666e-06,
+      "loss": 0.4445,
+      "step": 18467
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08945653098733e-06,
+      "loss": 0.4672,
+      "step": 18468
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08842508435264e-06,
+      "loss": 0.3931,
+      "step": 18469
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087393686847549e-06,
+      "loss": 0.4651,
+      "step": 18470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.086362338485012e-06,
+      "loss": 0.4288,
+      "step": 18471
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0853310392779865e-06,
+      "loss": 0.4409,
+      "step": 18472
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084299789239418e-06,
+      "loss": 0.4481,
+      "step": 18473
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.083268588382266e-06,
+      "loss": 0.4547,
+      "step": 18474
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.082237436719482e-06,
+      "loss": 0.5014,
+      "step": 18475
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0812063342640115e-06,
+      "loss": 0.4395,
+      "step": 18476
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080175281028809e-06,
+      "loss": 0.4191,
+      "step": 18477
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079144277026828e-06,
+      "loss": 0.4588,
+      "step": 18478
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078113322271008e-06,
+      "loss": 0.4106,
+      "step": 18479
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07708241677431e-06,
+      "loss": 0.45,
+      "step": 18480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076051560549674e-06,
+      "loss": 0.3614,
+      "step": 18481
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.075020753610052e-06,
+      "loss": 0.4463,
+      "step": 18482
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073989995968388e-06,
+      "loss": 0.4269,
+      "step": 18483
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07295928763763e-06,
+      "loss": 0.4733,
+      "step": 18484
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071928628630725e-06,
+      "loss": 0.4161,
+      "step": 18485
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0708980189606124e-06,
+      "loss": 0.4255,
+      "step": 18486
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069867458640242e-06,
+      "loss": 0.431,
+      "step": 18487
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068836947682559e-06,
+      "loss": 0.4562,
+      "step": 18488
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0678064861005e-06,
+      "loss": 0.482,
+      "step": 18489
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.066776073907015e-06,
+      "loss": 0.4243,
+      "step": 18490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0657457111150446e-06,
+      "loss": 0.4752,
+      "step": 18491
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064715397737524e-06,
+      "loss": 0.416,
+      "step": 18492
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063685133787398e-06,
+      "loss": 0.4354,
+      "step": 18493
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062654919277612e-06,
+      "loss": 0.4258,
+      "step": 18494
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.061624754221098e-06,
+      "loss": 0.4,
+      "step": 18495
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.060594638630797e-06,
+      "loss": 0.4361,
+      "step": 18496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05956457251965e-06,
+      "loss": 0.4277,
+      "step": 18497
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058534555900591e-06,
+      "loss": 0.4859,
+      "step": 18498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057504588786556e-06,
+      "loss": 0.4634,
+      "step": 18499
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05647467119049e-06,
+      "loss": 0.451,
+      "step": 18500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055444803125321e-06,
+      "loss": 0.4485,
+      "step": 18501
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054414984603982e-06,
+      "loss": 0.4554,
+      "step": 18502
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053385215639418e-06,
+      "loss": 0.4643,
+      "step": 18503
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052355496244553e-06,
+      "loss": 0.4346,
+      "step": 18504
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051325826432325e-06,
+      "loss": 0.4518,
+      "step": 18505
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050296206215667e-06,
+      "loss": 0.4938,
+      "step": 18506
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049266635607508e-06,
+      "loss": 0.4084,
+      "step": 18507
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048237114620781e-06,
+      "loss": 0.4063,
+      "step": 18508
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.04720764326842e-06,
+      "loss": 0.4194,
+      "step": 18509
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046178221563348e-06,
+      "loss": 0.4906,
+      "step": 18510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045148849518503e-06,
+      "loss": 0.4139,
+      "step": 18511
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.04411952714681e-06,
+      "loss": 0.4497,
+      "step": 18512
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043090254461194e-06,
+      "loss": 0.4463,
+      "step": 18513
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0420610314745865e-06,
+      "loss": 0.4215,
+      "step": 18514
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041031858199915e-06,
+      "loss": 0.475,
+      "step": 18515
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040002734650105e-06,
+      "loss": 0.4061,
+      "step": 18516
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038973660838083e-06,
+      "loss": 0.4525,
+      "step": 18517
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037944636776773e-06,
+      "loss": 0.472,
+      "step": 18518
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036915662479104e-06,
+      "loss": 0.4235,
+      "step": 18519
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.03588673795799e-06,
+      "loss": 0.4454,
+      "step": 18520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.034857863226364e-06,
+      "loss": 0.4486,
+      "step": 18521
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033829038297148e-06,
+      "loss": 0.4326,
+      "step": 18522
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032800263183256e-06,
+      "loss": 0.4371,
+      "step": 18523
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031771537897615e-06,
+      "loss": 0.4094,
+      "step": 18524
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.03074286245315e-06,
+      "loss": 0.4389,
+      "step": 18525
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029714236862774e-06,
+      "loss": 0.4267,
+      "step": 18526
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0286856611394106e-06,
+      "loss": 0.4587,
+      "step": 18527
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.02765713529598e-06,
+      "loss": 0.4388,
+      "step": 18528
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026628659345394e-06,
+      "loss": 0.4318,
+      "step": 18529
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0256002333005746e-06,
+      "loss": 0.4245,
+      "step": 18530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024571857174443e-06,
+      "loss": 0.4277,
+      "step": 18531
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.02354353097991e-06,
+      "loss": 0.4508,
+      "step": 18532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.02251525472989e-06,
+      "loss": 0.4614,
+      "step": 18533
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021487028437305e-06,
+      "loss": 0.4672,
+      "step": 18534
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020458852115063e-06,
+      "loss": 0.4066,
+      "step": 18535
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019430725776079e-06,
+      "loss": 0.4213,
+      "step": 18536
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018402649433274e-06,
+      "loss": 0.4236,
+      "step": 18537
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017374623099552e-06,
+      "loss": 0.3922,
+      "step": 18538
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.016346646787825e-06,
+      "loss": 0.4862,
+      "step": 18539
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0153187205110116e-06,
+      "loss": 0.4069,
+      "step": 18540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014290844282015e-06,
+      "loss": 0.4255,
+      "step": 18541
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01326301811375e-06,
+      "loss": 0.4629,
+      "step": 18542
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.012235242019125e-06,
+      "loss": 0.4395,
+      "step": 18543
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.011207516011049e-06,
+      "loss": 0.4504,
+      "step": 18544
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.010179840102426e-06,
+      "loss": 0.4646,
+      "step": 18545
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009152214306173e-06,
+      "loss": 0.4522,
+      "step": 18546
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008124638635189e-06,
+      "loss": 0.4511,
+      "step": 18547
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007097113102384e-06,
+      "loss": 0.4183,
+      "step": 18548
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.006069637720664e-06,
+      "loss": 0.4893,
+      "step": 18549
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00504221250293e-06,
+      "loss": 0.4754,
+      "step": 18550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00401483746209e-06,
+      "loss": 0.4304,
+      "step": 18551
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002987512611049e-06,
+      "loss": 0.4062,
+      "step": 18552
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001960237962709e-06,
+      "loss": 0.4349,
+      "step": 18553
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000933013529971e-06,
+      "loss": 0.3834,
+      "step": 18554
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999905839325737e-06,
+      "loss": 0.4548,
+      "step": 18555
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998878715362915e-06,
+      "loss": 0.4684,
+      "step": 18556
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.997851641654396e-06,
+      "loss": 0.4535,
+      "step": 18557
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996824618213084e-06,
+      "loss": 0.448,
+      "step": 18558
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995797645051884e-06,
+      "loss": 0.5296,
+      "step": 18559
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994770722183685e-06,
+      "loss": 0.4624,
+      "step": 18560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.993743849621389e-06,
+      "loss": 0.4153,
+      "step": 18561
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.992717027377899e-06,
+      "loss": 0.4028,
+      "step": 18562
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.991690255466105e-06,
+      "loss": 0.4244,
+      "step": 18563
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.990663533898906e-06,
+      "loss": 0.4294,
+      "step": 18564
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989636862689201e-06,
+      "loss": 0.4303,
+      "step": 18565
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988610241849877e-06,
+      "loss": 0.4319,
+      "step": 18566
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987583671393833e-06,
+      "loss": 0.4681,
+      "step": 18567
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986557151333967e-06,
+      "loss": 0.4252,
+      "step": 18568
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985530681683165e-06,
+      "loss": 0.4177,
+      "step": 18569
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.984504262454322e-06,
+      "loss": 0.4507,
+      "step": 18570
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983477893660333e-06,
+      "loss": 0.4367,
+      "step": 18571
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982451575314083e-06,
+      "loss": 0.4713,
+      "step": 18572
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981425307428467e-06,
+      "loss": 0.4009,
+      "step": 18573
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.980399090016378e-06,
+      "loss": 0.4377,
+      "step": 18574
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979372923090699e-06,
+      "loss": 0.4622,
+      "step": 18575
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.978346806664319e-06,
+      "loss": 0.3961,
+      "step": 18576
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977320740750133e-06,
+      "loss": 0.4026,
+      "step": 18577
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976294725361021e-06,
+      "loss": 0.3896,
+      "step": 18578
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975268760509871e-06,
+      "loss": 0.4352,
+      "step": 18579
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.974242846209574e-06,
+      "loss": 0.4454,
+      "step": 18580
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973216982473011e-06,
+      "loss": 0.4262,
+      "step": 18581
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9721911693130666e-06,
+      "loss": 0.4578,
+      "step": 18582
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.971165406742631e-06,
+      "loss": 0.4042,
+      "step": 18583
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9701396947745795e-06,
+      "loss": 0.467,
+      "step": 18584
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9691140334218e-06,
+      "loss": 0.4088,
+      "step": 18585
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.968088422697177e-06,
+      "loss": 0.4204,
+      "step": 18586
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967062862613585e-06,
+      "loss": 0.3707,
+      "step": 18587
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966037353183909e-06,
+      "loss": 0.4122,
+      "step": 18588
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965011894421032e-06,
+      "loss": 0.4835,
+      "step": 18589
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9639864863378335e-06,
+      "loss": 0.3887,
+      "step": 18590
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.962961128947188e-06,
+      "loss": 0.466,
+      "step": 18591
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961935822261975e-06,
+      "loss": 0.4184,
+      "step": 18592
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9609105662950796e-06,
+      "loss": 0.4254,
+      "step": 18593
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959885361059369e-06,
+      "loss": 0.4186,
+      "step": 18594
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958860206567727e-06,
+      "loss": 0.4375,
+      "step": 18595
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9578351028330285e-06,
+      "loss": 0.44,
+      "step": 18596
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956810049868143e-06,
+      "loss": 0.4173,
+      "step": 18597
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955785047685951e-06,
+      "loss": 0.4342,
+      "step": 18598
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.954760096299329e-06,
+      "loss": 0.4939,
+      "step": 18599
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9537351957211445e-06,
+      "loss": 0.4595,
+      "step": 18600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95271034596427e-06,
+      "loss": 0.4532,
+      "step": 18601
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.951685547041585e-06,
+      "loss": 0.4274,
+      "step": 18602
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.950660798965953e-06,
+      "loss": 0.4296,
+      "step": 18603
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949636101750247e-06,
+      "loss": 0.4741,
+      "step": 18604
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948611455407343e-06,
+      "loss": 0.4292,
+      "step": 18605
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947586859950103e-06,
+      "loss": 0.4494,
+      "step": 18606
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9465623153913976e-06,
+      "loss": 0.4902,
+      "step": 18607
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.945537821744101e-06,
+      "loss": 0.4567,
+      "step": 18608
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944513379021072e-06,
+      "loss": 0.4745,
+      "step": 18609
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9434889872351845e-06,
+      "loss": 0.4235,
+      "step": 18610
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942464646399302e-06,
+      "loss": 0.4804,
+      "step": 18611
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941440356526291e-06,
+      "loss": 0.485,
+      "step": 18612
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940416117629014e-06,
+      "loss": 0.4133,
+      "step": 18613
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9393919297203415e-06,
+      "loss": 0.4594,
+      "step": 18614
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.938367792813131e-06,
+      "loss": 0.4601,
+      "step": 18615
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937343706920251e-06,
+      "loss": 0.4545,
+      "step": 18616
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936319672054562e-06,
+      "loss": 0.4459,
+      "step": 18617
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935295688228923e-06,
+      "loss": 0.4656,
+      "step": 18618
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.934271755456196e-06,
+      "loss": 0.4752,
+      "step": 18619
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933247873749249e-06,
+      "loss": 0.4834,
+      "step": 18620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932224043120931e-06,
+      "loss": 0.4439,
+      "step": 18621
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93120026358411e-06,
+      "loss": 0.4526,
+      "step": 18622
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930176535151644e-06,
+      "loss": 0.4479,
+      "step": 18623
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929152857836384e-06,
+      "loss": 0.4375,
+      "step": 18624
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.928129231651192e-06,
+      "loss": 0.4507,
+      "step": 18625
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.927105656608927e-06,
+      "loss": 0.4107,
+      "step": 18626
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.926082132722445e-06,
+      "loss": 0.395,
+      "step": 18627
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.925058660004594e-06,
+      "loss": 0.4053,
+      "step": 18628
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.924035238468238e-06,
+      "loss": 0.4267,
+      "step": 18629
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923011868126229e-06,
+      "loss": 0.4288,
+      "step": 18630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.921988548991415e-06,
+      "loss": 0.4279,
+      "step": 18631
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920965281076657e-06,
+      "loss": 0.4583,
+      "step": 18632
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919942064394805e-06,
+      "loss": 0.4303,
+      "step": 18633
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.918918898958706e-06,
+      "loss": 0.4576,
+      "step": 18634
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917895784781211e-06,
+      "loss": 0.4315,
+      "step": 18635
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916872721875181e-06,
+      "loss": 0.4509,
+      "step": 18636
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.915849710253454e-06,
+      "loss": 0.4203,
+      "step": 18637
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9148267499288835e-06,
+      "loss": 0.3885,
+      "step": 18638
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.913803840914321e-06,
+      "loss": 0.4207,
+      "step": 18639
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.912780983222606e-06,
+      "loss": 0.4721,
+      "step": 18640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.911758176866591e-06,
+      "loss": 0.4512,
+      "step": 18641
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910735421859127e-06,
+      "loss": 0.4087,
+      "step": 18642
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9097127182130535e-06,
+      "loss": 0.405,
+      "step": 18643
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908690065941214e-06,
+      "loss": 0.4824,
+      "step": 18644
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.907667465056462e-06,
+      "loss": 0.4377,
+      "step": 18645
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906644915571631e-06,
+      "loss": 0.4873,
+      "step": 18646
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90562241749957e-06,
+      "loss": 0.4045,
+      "step": 18647
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904599970853122e-06,
+      "loss": 0.436,
+      "step": 18648
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9035775756451284e-06,
+      "loss": 0.4301,
+      "step": 18649
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.902555231888427e-06,
+      "loss": 0.4287,
+      "step": 18650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901532939595867e-06,
+      "loss": 0.4041,
+      "step": 18651
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900510698780275e-06,
+      "loss": 0.4466,
+      "step": 18652
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.899488509454503e-06,
+      "loss": 0.4498,
+      "step": 18653
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8984663716313864e-06,
+      "loss": 0.3993,
+      "step": 18654
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897444285323759e-06,
+      "loss": 0.4678,
+      "step": 18655
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89642225054446e-06,
+      "loss": 0.4735,
+      "step": 18656
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89540026730633e-06,
+      "loss": 0.4064,
+      "step": 18657
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.894378335622202e-06,
+      "loss": 0.4595,
+      "step": 18658
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893356455504911e-06,
+      "loss": 0.4349,
+      "step": 18659
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892334626967295e-06,
+      "loss": 0.4137,
+      "step": 18660
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891312850022184e-06,
+      "loss": 0.4287,
+      "step": 18661
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.890291124682413e-06,
+      "loss": 0.4342,
+      "step": 18662
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889269450960817e-06,
+      "loss": 0.4678,
+      "step": 18663
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.888247828870231e-06,
+      "loss": 0.4083,
+      "step": 18664
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887226258423475e-06,
+      "loss": 0.4581,
+      "step": 18665
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.88620473963339e-06,
+      "loss": 0.4638,
+      "step": 18666
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885183272512808e-06,
+      "loss": 0.4528,
+      "step": 18667
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8841618570745494e-06,
+      "loss": 0.4419,
+      "step": 18668
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883140493331451e-06,
+      "loss": 0.4466,
+      "step": 18669
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8821191812963396e-06,
+      "loss": 0.4374,
+      "step": 18670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.881097920982039e-06,
+      "loss": 0.4182,
+      "step": 18671
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880076712401379e-06,
+      "loss": 0.4377,
+      "step": 18672
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87905555556719e-06,
+      "loss": 0.4236,
+      "step": 18673
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878034450492293e-06,
+      "loss": 0.4299,
+      "step": 18674
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87701339718951e-06,
+      "loss": 0.4573,
+      "step": 18675
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8759923956716766e-06,
+      "loss": 0.4713,
+      "step": 18676
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8749714459516036e-06,
+      "loss": 0.4728,
+      "step": 18677
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87395054804212e-06,
+      "loss": 0.4028,
+      "step": 18678
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.872929701956054e-06,
+      "loss": 0.4513,
+      "step": 18679
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8719089077062195e-06,
+      "loss": 0.4154,
+      "step": 18680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8708881653054385e-06,
+      "loss": 0.4515,
+      "step": 18681
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869867474766538e-06,
+      "loss": 0.4053,
+      "step": 18682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868846836102328e-06,
+      "loss": 0.4248,
+      "step": 18683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867826249325634e-06,
+      "loss": 0.3744,
+      "step": 18684
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866805714449276e-06,
+      "loss": 0.4616,
+      "step": 18685
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.86578523148607e-06,
+      "loss": 0.4166,
+      "step": 18686
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8647648004488304e-06,
+      "loss": 0.3931,
+      "step": 18687
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.863744421350379e-06,
+      "loss": 0.4661,
+      "step": 18688
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862724094203527e-06,
+      "loss": 0.4264,
+      "step": 18689
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861703819021094e-06,
+      "loss": 0.3864,
+      "step": 18690
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.860683595815894e-06,
+      "loss": 0.4632,
+      "step": 18691
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859663424600736e-06,
+      "loss": 0.429,
+      "step": 18692
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.858643305388438e-06,
+      "loss": 0.4571,
+      "step": 18693
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.857623238191815e-06,
+      "loss": 0.4506,
+      "step": 18694
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856603223023673e-06,
+      "loss": 0.46,
+      "step": 18695
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.855583259896829e-06,
+      "loss": 0.4495,
+      "step": 18696
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.854563348824091e-06,
+      "loss": 0.4676,
+      "step": 18697
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853543489818267e-06,
+      "loss": 0.4036,
+      "step": 18698
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852523682892171e-06,
+      "loss": 0.4573,
+      "step": 18699
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.851503928058608e-06,
+      "loss": 0.4642,
+      "step": 18700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850484225330393e-06,
+      "loss": 0.4689,
+      "step": 18701
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849464574720324e-06,
+      "loss": 0.4709,
+      "step": 18702
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848444976241212e-06,
+      "loss": 0.4218,
+      "step": 18703
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847425429905866e-06,
+      "loss": 0.4453,
+      "step": 18704
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846405935727088e-06,
+      "loss": 0.4757,
+      "step": 18705
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.845386493717684e-06,
+      "loss": 0.4267,
+      "step": 18706
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844367103890461e-06,
+      "loss": 0.4336,
+      "step": 18707
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843347766258216e-06,
+      "loss": 0.4323,
+      "step": 18708
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8423284808337565e-06,
+      "loss": 0.4192,
+      "step": 18709
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841309247629887e-06,
+      "loss": 0.4036,
+      "step": 18710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840290066659405e-06,
+      "loss": 0.3944,
+      "step": 18711
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.839270937935111e-06,
+      "loss": 0.4633,
+      "step": 18712
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8382518614698105e-06,
+      "loss": 0.4486,
+      "step": 18713
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8372328372762956e-06,
+      "loss": 0.4488,
+      "step": 18714
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.836213865367369e-06,
+      "loss": 0.4434,
+      "step": 18715
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.835194945755834e-06,
+      "loss": 0.4089,
+      "step": 18716
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8341760784544796e-06,
+      "loss": 0.4655,
+      "step": 18717
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833157263476111e-06,
+      "loss": 0.4053,
+      "step": 18718
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832138500833518e-06,
+      "loss": 0.473,
+      "step": 18719
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8311197905395015e-06,
+      "loss": 0.4136,
+      "step": 18720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830101132606849e-06,
+      "loss": 0.4275,
+      "step": 18721
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.829082527048364e-06,
+      "loss": 0.4329,
+      "step": 18722
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.828063973876834e-06,
+      "loss": 0.4268,
+      "step": 18723
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827045473105053e-06,
+      "loss": 0.4447,
+      "step": 18724
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.826027024745817e-06,
+      "loss": 0.4644,
+      "step": 18725
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825008628811914e-06,
+      "loss": 0.3893,
+      "step": 18726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.823990285316135e-06,
+      "loss": 0.3985,
+      "step": 18727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822971994271276e-06,
+      "loss": 0.4919,
+      "step": 18728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82195375569012e-06,
+      "loss": 0.4716,
+      "step": 18729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.820935569585461e-06,
+      "loss": 0.4255,
+      "step": 18730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819917435970086e-06,
+      "loss": 0.3748,
+      "step": 18731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8188993548567775e-06,
+      "loss": 0.4641,
+      "step": 18732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817881326258328e-06,
+      "loss": 0.4568,
+      "step": 18733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8168633501875225e-06,
+      "loss": 0.4293,
+      "step": 18734
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815845426657153e-06,
+      "loss": 0.3942,
+      "step": 18735
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.814827555679994e-06,
+      "loss": 0.4974,
+      "step": 18736
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.813809737268837e-06,
+      "loss": 0.4015,
+      "step": 18737
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.812791971436467e-06,
+      "loss": 0.4388,
+      "step": 18738
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811774258195659e-06,
+      "loss": 0.4274,
+      "step": 18739
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.810756597559206e-06,
+      "loss": 0.4055,
+      "step": 18740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809738989539886e-06,
+      "loss": 0.4988,
+      "step": 18741
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808721434150474e-06,
+      "loss": 0.4321,
+      "step": 18742
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8077039314037555e-06,
+      "loss": 0.4045,
+      "step": 18743
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806686481312513e-06,
+      "loss": 0.3944,
+      "step": 18744
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.805669083889522e-06,
+      "loss": 0.4766,
+      "step": 18745
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.80465173914756e-06,
+      "loss": 0.3744,
+      "step": 18746
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803634447099412e-06,
+      "loss": 0.3898,
+      "step": 18747
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802617207757846e-06,
+      "loss": 0.4167,
+      "step": 18748
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801600021135643e-06,
+      "loss": 0.4527,
+      "step": 18749
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.800582887245584e-06,
+      "loss": 0.4328,
+      "step": 18750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799565806100438e-06,
+      "loss": 0.4404,
+      "step": 18751
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798548777712977e-06,
+      "loss": 0.4593,
+      "step": 18752
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7975318020959815e-06,
+      "loss": 0.4603,
+      "step": 18753
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796514879262219e-06,
+      "loss": 0.4457,
+      "step": 18754
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795498009224465e-06,
+      "loss": 0.4036,
+      "step": 18755
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794481191995496e-06,
+      "loss": 0.4358,
+      "step": 18756
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793464427588075e-06,
+      "loss": 0.4427,
+      "step": 18757
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.792447716014975e-06,
+      "loss": 0.4145,
+      "step": 18758
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.791431057288972e-06,
+      "loss": 0.4358,
+      "step": 18759
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.790414451422827e-06,
+      "loss": 0.4074,
+      "step": 18760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789397898429317e-06,
+      "loss": 0.4339,
+      "step": 18761
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788381398321203e-06,
+      "loss": 0.4368,
+      "step": 18762
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787364951111252e-06,
+      "loss": 0.5003,
+      "step": 18763
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786348556812231e-06,
+      "loss": 0.4424,
+      "step": 18764
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785332215436915e-06,
+      "loss": 0.4387,
+      "step": 18765
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784315926998057e-06,
+      "loss": 0.3983,
+      "step": 18766
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.783299691508428e-06,
+      "loss": 0.4399,
+      "step": 18767
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782283508980793e-06,
+      "loss": 0.4456,
+      "step": 18768
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781267379427912e-06,
+      "loss": 0.4543,
+      "step": 18769
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780251302862547e-06,
+      "loss": 0.4208,
+      "step": 18770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779235279297467e-06,
+      "loss": 0.4168,
+      "step": 18771
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.778219308745427e-06,
+      "loss": 0.4577,
+      "step": 18772
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.777203391219187e-06,
+      "loss": 0.462,
+      "step": 18773
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776187526731507e-06,
+      "loss": 0.4662,
+      "step": 18774
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.775171715295154e-06,
+      "loss": 0.4455,
+      "step": 18775
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774155956922876e-06,
+      "loss": 0.3685,
+      "step": 18776
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773140251627436e-06,
+      "loss": 0.4247,
+      "step": 18777
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.772124599421595e-06,
+      "loss": 0.4543,
+      "step": 18778
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771109000318102e-06,
+      "loss": 0.417,
+      "step": 18779
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.770093454329716e-06,
+      "loss": 0.4775,
+      "step": 18780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.769077961469197e-06,
+      "loss": 0.4371,
+      "step": 18781
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.768062521749297e-06,
+      "loss": 0.4475,
+      "step": 18782
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.767047135182764e-06,
+      "loss": 0.4684,
+      "step": 18783
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76603180178236e-06,
+      "loss": 0.4651,
+      "step": 18784
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76501652156083e-06,
+      "loss": 0.4747,
+      "step": 18785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76400129453093e-06,
+      "loss": 0.4438,
+      "step": 18786
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7629861207054135e-06,
+      "loss": 0.4349,
+      "step": 18787
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761971000097027e-06,
+      "loss": 0.4483,
+      "step": 18788
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7609559327185205e-06,
+      "loss": 0.4336,
+      "step": 18789
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7599409185826495e-06,
+      "loss": 0.4389,
+      "step": 18790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758925957702153e-06,
+      "loss": 0.3778,
+      "step": 18791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7579110500897885e-06,
+      "loss": 0.3834,
+      "step": 18792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756896195758299e-06,
+      "loss": 0.4512,
+      "step": 18793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755881394720427e-06,
+      "loss": 0.4844,
+      "step": 18794
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754866646988924e-06,
+      "loss": 0.4566,
+      "step": 18795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753851952576535e-06,
+      "loss": 0.4662,
+      "step": 18796
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752837311496e-06,
+      "loss": 0.4339,
+      "step": 18797
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7518227237600665e-06,
+      "loss": 0.4187,
+      "step": 18798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.750808189381482e-06,
+      "loss": 0.5248,
+      "step": 18799
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.749793708372979e-06,
+      "loss": 0.41,
+      "step": 18800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748779280747308e-06,
+      "loss": 0.4169,
+      "step": 18801
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747764906517208e-06,
+      "loss": 0.4327,
+      "step": 18802
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746750585695421e-06,
+      "loss": 0.4442,
+      "step": 18803
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.74573631829468e-06,
+      "loss": 0.4401,
+      "step": 18804
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744722104327733e-06,
+      "loss": 0.4085,
+      "step": 18805
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743707943807312e-06,
+      "loss": 0.4311,
+      "step": 18806
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742693836746156e-06,
+      "loss": 0.4527,
+      "step": 18807
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741679783157005e-06,
+      "loss": 0.4134,
+      "step": 18808
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740665783052598e-06,
+      "loss": 0.4132,
+      "step": 18809
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7396518364456635e-06,
+      "loss": 0.4665,
+      "step": 18810
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73863794334894e-06,
+      "loss": 0.3999,
+      "step": 18811
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.737624103775168e-06,
+      "loss": 0.4009,
+      "step": 18812
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736610317737075e-06,
+      "loss": 0.4483,
+      "step": 18813
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735596585247389e-06,
+      "loss": 0.4735,
+      "step": 18814
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.734582906318855e-06,
+      "loss": 0.41,
+      "step": 18815
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733569280964195e-06,
+      "loss": 0.4396,
+      "step": 18816
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732555709196143e-06,
+      "loss": 0.4181,
+      "step": 18817
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731542191027435e-06,
+      "loss": 0.4107,
+      "step": 18818
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730528726470792e-06,
+      "loss": 0.4192,
+      "step": 18819
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729515315538947e-06,
+      "loss": 0.4444,
+      "step": 18820
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.728501958244632e-06,
+      "loss": 0.4336,
+      "step": 18821
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727488654600568e-06,
+      "loss": 0.404,
+      "step": 18822
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.726475404619491e-06,
+      "loss": 0.4309,
+      "step": 18823
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.725462208314117e-06,
+      "loss": 0.4035,
+      "step": 18824
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.724449065697182e-06,
+      "loss": 0.44,
+      "step": 18825
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.723435976781402e-06,
+      "loss": 0.4457,
+      "step": 18826
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722422941579508e-06,
+      "loss": 0.4328,
+      "step": 18827
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7214099601042195e-06,
+      "loss": 0.4226,
+      "step": 18828
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720397032368261e-06,
+      "loss": 0.4517,
+      "step": 18829
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.719384158384359e-06,
+      "loss": 0.4159,
+      "step": 18830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7183713381652275e-06,
+      "loss": 0.4277,
+      "step": 18831
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7173585717235926e-06,
+      "loss": 0.4131,
+      "step": 18832
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716345859072178e-06,
+      "loss": 0.4264,
+      "step": 18833
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.715333200223694e-06,
+      "loss": 0.4742,
+      "step": 18834
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7143205951908695e-06,
+      "loss": 0.4847,
+      "step": 18835
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7133080439864195e-06,
+      "loss": 0.4359,
+      "step": 18836
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7122955466230555e-06,
+      "loss": 0.4334,
+      "step": 18837
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711283103113501e-06,
+      "loss": 0.4719,
+      "step": 18838
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710270713470473e-06,
+      "loss": 0.4475,
+      "step": 18839
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.709258377706681e-06,
+      "loss": 0.4704,
+      "step": 18840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7082460958348465e-06,
+      "loss": 0.5016,
+      "step": 18841
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.707233867867683e-06,
+      "loss": 0.3996,
+      "step": 18842
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7062216938179e-06,
+      "loss": 0.4317,
+      "step": 18843
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705209573698212e-06,
+      "loss": 0.4398,
+      "step": 18844
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704197507521336e-06,
+      "loss": 0.4639,
+      "step": 18845
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.703185495299981e-06,
+      "loss": 0.4183,
+      "step": 18846
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702173537046851e-06,
+      "loss": 0.4536,
+      "step": 18847
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7011616327746645e-06,
+      "loss": 0.4146,
+      "step": 18848
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700149782496132e-06,
+      "loss": 0.4721,
+      "step": 18849
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699137986223955e-06,
+      "loss": 0.4642,
+      "step": 18850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698126243970845e-06,
+      "loss": 0.4675,
+      "step": 18851
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6971145557495145e-06,
+      "loss": 0.4231,
+      "step": 18852
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6961029215726615e-06,
+      "loss": 0.4358,
+      "step": 18853
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695091341452999e-06,
+      "loss": 0.4939,
+      "step": 18854
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694079815403233e-06,
+      "loss": 0.4939,
+      "step": 18855
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.693068343436066e-06,
+      "loss": 0.4088,
+      "step": 18856
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692056925564196e-06,
+      "loss": 0.4531,
+      "step": 18857
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.691045561800339e-06,
+      "loss": 0.4462,
+      "step": 18858
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690034252157186e-06,
+      "loss": 0.4159,
+      "step": 18859
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689022996647444e-06,
+      "loss": 0.4725,
+      "step": 18860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688011795283819e-06,
+      "loss": 0.4223,
+      "step": 18861
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687000648079002e-06,
+      "loss": 0.3658,
+      "step": 18862
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6859895550457e-06,
+      "loss": 0.4535,
+      "step": 18863
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.684978516196614e-06,
+      "loss": 0.4074,
+      "step": 18864
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6839675315444365e-06,
+      "loss": 0.3992,
+      "step": 18865
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682956601101872e-06,
+      "loss": 0.4866,
+      "step": 18866
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681945724881613e-06,
+      "loss": 0.4313,
+      "step": 18867
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6809349028963555e-06,
+      "loss": 0.4487,
+      "step": 18868
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679924135158796e-06,
+      "loss": 0.4568,
+      "step": 18869
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6789134216816365e-06,
+      "loss": 0.4,
+      "step": 18870
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677902762477563e-06,
+      "loss": 0.4474,
+      "step": 18871
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6768921575592705e-06,
+      "loss": 0.4228,
+      "step": 18872
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675881606939461e-06,
+      "loss": 0.4649,
+      "step": 18873
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674871110630815e-06,
+      "loss": 0.4528,
+      "step": 18874
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67386066864603e-06,
+      "loss": 0.3907,
+      "step": 18875
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6728502809978035e-06,
+      "loss": 0.4524,
+      "step": 18876
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671839947698818e-06,
+      "loss": 0.444,
+      "step": 18877
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670829668761762e-06,
+      "loss": 0.3841,
+      "step": 18878
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669819444199331e-06,
+      "loss": 0.4937,
+      "step": 18879
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668809274024206e-06,
+      "loss": 0.4619,
+      "step": 18880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66779915824908e-06,
+      "loss": 0.3965,
+      "step": 18881
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666789096886639e-06,
+      "loss": 0.4203,
+      "step": 18882
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.665779089949574e-06,
+      "loss": 0.4311,
+      "step": 18883
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664769137450561e-06,
+      "loss": 0.4463,
+      "step": 18884
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6637592394022915e-06,
+      "loss": 0.4749,
+      "step": 18885
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662749395817453e-06,
+      "loss": 0.4247,
+      "step": 18886
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6617396067087246e-06,
+      "loss": 0.4018,
+      "step": 18887
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660729872088786e-06,
+      "loss": 0.4286,
+      "step": 18888
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659720191970327e-06,
+      "loss": 0.4386,
+      "step": 18889
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.658710566366022e-06,
+      "loss": 0.4332,
+      "step": 18890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657700995288554e-06,
+      "loss": 0.4236,
+      "step": 18891
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65669147875061e-06,
+      "loss": 0.4551,
+      "step": 18892
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6556820167648605e-06,
+      "loss": 0.4403,
+      "step": 18893
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654672609343987e-06,
+      "loss": 0.4379,
+      "step": 18894
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.653663256500673e-06,
+      "loss": 0.4579,
+      "step": 18895
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65265395824759e-06,
+      "loss": 0.4406,
+      "step": 18896
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651644714597417e-06,
+      "loss": 0.4781,
+      "step": 18897
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.650635525562829e-06,
+      "loss": 0.4156,
+      "step": 18898
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.649626391156506e-06,
+      "loss": 0.4512,
+      "step": 18899
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6486173113911135e-06,
+      "loss": 0.4346,
+      "step": 18900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647608286279338e-06,
+      "loss": 0.471,
+      "step": 18901
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646599315833841e-06,
+      "loss": 0.4604,
+      "step": 18902
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645590400067299e-06,
+      "loss": 0.4473,
+      "step": 18903
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6445815389923904e-06,
+      "loss": 0.4042,
+      "step": 18904
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643572732621778e-06,
+      "loss": 0.4426,
+      "step": 18905
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642563980968135e-06,
+      "loss": 0.4311,
+      "step": 18906
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6415552840441375e-06,
+      "loss": 0.4965,
+      "step": 18907
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.640546641862445e-06,
+      "loss": 0.488,
+      "step": 18908
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639538054435736e-06,
+      "loss": 0.4766,
+      "step": 18909
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638529521776673e-06,
+      "loss": 0.4598,
+      "step": 18910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637521043897918e-06,
+      "loss": 0.4468,
+      "step": 18911
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636512620812144e-06,
+      "loss": 0.4631,
+      "step": 18912
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.63550425253202e-06,
+      "loss": 0.4399,
+      "step": 18913
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.634495939070203e-06,
+      "loss": 0.4347,
+      "step": 18914
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633487680439362e-06,
+      "loss": 0.4017,
+      "step": 18915
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632479476652163e-06,
+      "loss": 0.4995,
+      "step": 18916
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631471327721263e-06,
+      "loss": 0.4052,
+      "step": 18917
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6304632336593275e-06,
+      "loss": 0.4269,
+      "step": 18918
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629455194479023e-06,
+      "loss": 0.5271,
+      "step": 18919
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.628447210193007e-06,
+      "loss": 0.4331,
+      "step": 18920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627439280813933e-06,
+      "loss": 0.4551,
+      "step": 18921
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626431406354467e-06,
+      "loss": 0.4389,
+      "step": 18922
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.625423586827272e-06,
+      "loss": 0.4431,
+      "step": 18923
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624415822244997e-06,
+      "loss": 0.4937,
+      "step": 18924
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623408112620305e-06,
+      "loss": 0.4403,
+      "step": 18925
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622400457965856e-06,
+      "loss": 0.4529,
+      "step": 18926
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621392858294299e-06,
+      "loss": 0.4411,
+      "step": 18927
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620385313618292e-06,
+      "loss": 0.4551,
+      "step": 18928
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.619377823950495e-06,
+      "loss": 0.4331,
+      "step": 18929
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.618370389303558e-06,
+      "loss": 0.4299,
+      "step": 18930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.617363009690131e-06,
+      "loss": 0.4394,
+      "step": 18931
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6163556851228725e-06,
+      "loss": 0.4264,
+      "step": 18932
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61534841561443e-06,
+      "loss": 0.4499,
+      "step": 18933
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6143412011774556e-06,
+      "loss": 0.4563,
+      "step": 18934
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6133340418246055e-06,
+      "loss": 0.4824,
+      "step": 18935
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612326937568522e-06,
+      "loss": 0.4214,
+      "step": 18936
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611319888421858e-06,
+      "loss": 0.4518,
+      "step": 18937
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610312894397265e-06,
+      "loss": 0.4575,
+      "step": 18938
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609305955507385e-06,
+      "loss": 0.4443,
+      "step": 18939
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6082990717648715e-06,
+      "loss": 0.4278,
+      "step": 18940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6072922431823675e-06,
+      "loss": 0.4386,
+      "step": 18941
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.606285469772514e-06,
+      "loss": 0.4477,
+      "step": 18942
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6052787515479634e-06,
+      "loss": 0.4438,
+      "step": 18943
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604272088521361e-06,
+      "loss": 0.4664,
+      "step": 18944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6032654807053425e-06,
+      "loss": 0.4394,
+      "step": 18945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.602258928112556e-06,
+      "loss": 0.4027,
+      "step": 18946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601252430755648e-06,
+      "loss": 0.3967,
+      "step": 18947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.600245988647252e-06,
+      "loss": 0.4143,
+      "step": 18948
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599239601800013e-06,
+      "loss": 0.454,
+      "step": 18949
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598233270226574e-06,
+      "loss": 0.4429,
+      "step": 18950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5972269939395725e-06,
+      "loss": 0.46,
+      "step": 18951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596220772951643e-06,
+      "loss": 0.4266,
+      "step": 18952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.595214607275425e-06,
+      "loss": 0.4439,
+      "step": 18953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594208496923566e-06,
+      "loss": 0.4452,
+      "step": 18954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5932024419086876e-06,
+      "loss": 0.4873,
+      "step": 18955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592196442243435e-06,
+      "loss": 0.4479,
+      "step": 18956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.591190497940446e-06,
+      "loss": 0.4669,
+      "step": 18957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590184609012345e-06,
+      "loss": 0.4996,
+      "step": 18958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589178775471775e-06,
+      "loss": 0.4377,
+      "step": 18959
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588172997331371e-06,
+      "loss": 0.4907,
+      "step": 18960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.58716727460376e-06,
+      "loss": 0.3733,
+      "step": 18961
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.58616160730157e-06,
+      "loss": 0.4467,
+      "step": 18962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5851559954374434e-06,
+      "loss": 0.4033,
+      "step": 18963
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.584150439023998e-06,
+      "loss": 0.4058,
+      "step": 18964
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583144938073874e-06,
+      "loss": 0.4154,
+      "step": 18965
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5821394925996985e-06,
+      "loss": 0.4599,
+      "step": 18966
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.581134102614096e-06,
+      "loss": 0.4211,
+      "step": 18967
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5801287681296955e-06,
+      "loss": 0.4872,
+      "step": 18968
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579123489159129e-06,
+      "loss": 0.4365,
+      "step": 18969
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578118265715016e-06,
+      "loss": 0.5236,
+      "step": 18970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.4189,
+      "step": 18971
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576107985456669e-06,
+      "loss": 0.4757,
+      "step": 18972
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.575102928667677e-06,
+      "loss": 0.4597,
+      "step": 18973
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.57409792745564e-06,
+      "loss": 0.452,
+      "step": 18974
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573092981833184e-06,
+      "loss": 0.4351,
+      "step": 18975
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572088091812924e-06,
+      "loss": 0.4286,
+      "step": 18976
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.571083257407486e-06,
+      "loss": 0.3846,
+      "step": 18977
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570078478629495e-06,
+      "loss": 0.4782,
+      "step": 18978
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.569073755491561e-06,
+      "loss": 0.4508,
+      "step": 18979
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.56806908800631e-06,
+      "loss": 0.4511,
+      "step": 18980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567064476186363e-06,
+      "loss": 0.3919,
+      "step": 18981
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566059920044336e-06,
+      "loss": 0.4214,
+      "step": 18982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565055419592837e-06,
+      "loss": 0.4595,
+      "step": 18983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564050974844497e-06,
+      "loss": 0.4113,
+      "step": 18984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.563046585811922e-06,
+      "loss": 0.4186,
+      "step": 18985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562042252507729e-06,
+      "loss": 0.474,
+      "step": 18986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561037974944538e-06,
+      "loss": 0.4552,
+      "step": 18987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5600337531349544e-06,
+      "loss": 0.4436,
+      "step": 18988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.559029587091596e-06,
+      "loss": 0.4488,
+      "step": 18989
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558025476827075e-06,
+      "loss": 0.4144,
+      "step": 18990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.557021422354005e-06,
+      "loss": 0.4267,
+      "step": 18991
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5560174236849964e-06,
+      "loss": 0.4307,
+      "step": 18992
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555013480832654e-06,
+      "loss": 0.4091,
+      "step": 18993
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.554009593809595e-06,
+      "loss": 0.4257,
+      "step": 18994
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553005762628421e-06,
+      "loss": 0.4795,
+      "step": 18995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552001987301743e-06,
+      "loss": 0.4599,
+      "step": 18996
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5509982678421715e-06,
+      "loss": 0.3841,
+      "step": 18997
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.54999460426231e-06,
+      "loss": 0.3835,
+      "step": 18998
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548990996574763e-06,
+      "loss": 0.4921,
+      "step": 18999
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547987444792143e-06,
+      "loss": 0.4081,
+      "step": 19000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546983948927047e-06,
+      "loss": 0.4426,
+      "step": 19001
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545980508992086e-06,
+      "loss": 0.4244,
+      "step": 19002
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.544977124999854e-06,
+      "loss": 0.4773,
+      "step": 19003
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543973796962964e-06,
+      "loss": 0.4487,
+      "step": 19004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542970524894008e-06,
+      "loss": 0.4409,
+      "step": 19005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541967308805596e-06,
+      "loss": 0.3759,
+      "step": 19006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.540964148710319e-06,
+      "loss": 0.4241,
+      "step": 19007
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539961044620785e-06,
+      "loss": 0.3947,
+      "step": 19008
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538957996549592e-06,
+      "loss": 0.4541,
+      "step": 19009
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.537955004509332e-06,
+      "loss": 0.4664,
+      "step": 19010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.536952068512609e-06,
+      "loss": 0.4456,
+      "step": 19011
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.535949188572021e-06,
+      "loss": 0.4177,
+      "step": 19012
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.534946364700156e-06,
+      "loss": 0.4,
+      "step": 19013
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53394359690962e-06,
+      "loss": 0.4373,
+      "step": 19014
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5329408852130025e-06,
+      "loss": 0.4271,
+      "step": 19015
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531938229622893e-06,
+      "loss": 0.4257,
+      "step": 19016
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5309356301518905e-06,
+      "loss": 0.4732,
+      "step": 19017
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52993308681259e-06,
+      "loss": 0.4222,
+      "step": 19018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528930599617577e-06,
+      "loss": 0.4935,
+      "step": 19019
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5279281685794465e-06,
+      "loss": 0.4114,
+      "step": 19020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526925793710791e-06,
+      "loss": 0.4596,
+      "step": 19021
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525923475024195e-06,
+      "loss": 0.4676,
+      "step": 19022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.524921212532251e-06,
+      "loss": 0.4229,
+      "step": 19023
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52391900624755e-06,
+      "loss": 0.4466,
+      "step": 19024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.522916856182678e-06,
+      "loss": 0.4658,
+      "step": 19025
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.521914762350216e-06,
+      "loss": 0.44,
+      "step": 19026
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520912724762757e-06,
+      "loss": 0.4566,
+      "step": 19027
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.519910743432888e-06,
+      "loss": 0.4429,
+      "step": 19028
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.518908818373188e-06,
+      "loss": 0.4633,
+      "step": 19029
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517906949596243e-06,
+      "loss": 0.4409,
+      "step": 19030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516905137114642e-06,
+      "loss": 0.4218,
+      "step": 19031
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.51590338094096e-06,
+      "loss": 0.441,
+      "step": 19032
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514901681087783e-06,
+      "loss": 0.4217,
+      "step": 19033
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.513900037567694e-06,
+      "loss": 0.4358,
+      "step": 19034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5128984503932734e-06,
+      "loss": 0.4759,
+      "step": 19035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5118969195770955e-06,
+      "loss": 0.4404,
+      "step": 19036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510895445131748e-06,
+      "loss": 0.4059,
+      "step": 19037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509894027069801e-06,
+      "loss": 0.4345,
+      "step": 19038
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5088926654038355e-06,
+      "loss": 0.4222,
+      "step": 19039
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.507891360146435e-06,
+      "loss": 0.4902,
+      "step": 19040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.506890111310166e-06,
+      "loss": 0.435,
+      "step": 19041
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505888918907609e-06,
+      "loss": 0.4088,
+      "step": 19042
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.504887782951343e-06,
+      "loss": 0.4466,
+      "step": 19043
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503886703453933e-06,
+      "loss": 0.4191,
+      "step": 19044
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5028856804279626e-06,
+      "loss": 0.4344,
+      "step": 19045
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501884713886001e-06,
+      "loss": 0.4987,
+      "step": 19046
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500883803840613e-06,
+      "loss": 0.4598,
+      "step": 19047
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.499882950304379e-06,
+      "loss": 0.4415,
+      "step": 19048
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49888215328987e-06,
+      "loss": 0.4306,
+      "step": 19049
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.497881412809649e-06,
+      "loss": 0.4632,
+      "step": 19050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49688072887629e-06,
+      "loss": 0.467,
+      "step": 19051
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495880101502364e-06,
+      "loss": 0.4332,
+      "step": 19052
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494879530700434e-06,
+      "loss": 0.4104,
+      "step": 19053
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.493879016483068e-06,
+      "loss": 0.3779,
+      "step": 19054
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4928785588628385e-06,
+      "loss": 0.5036,
+      "step": 19055
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.491878157852308e-06,
+      "loss": 0.4671,
+      "step": 19056
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4908778134640336e-06,
+      "loss": 0.3972,
+      "step": 19057
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4898775257105914e-06,
+      "loss": 0.4723,
+      "step": 19058
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488877294604538e-06,
+      "loss": 0.4585,
+      "step": 19059
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.487877120158437e-06,
+      "loss": 0.4311,
+      "step": 19060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486877002384856e-06,
+      "loss": 0.4281,
+      "step": 19061
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.485876941296348e-06,
+      "loss": 0.4533,
+      "step": 19062
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484876936905479e-06,
+      "loss": 0.4193,
+      "step": 19063
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483876989224809e-06,
+      "loss": 0.4746,
+      "step": 19064
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.482877098266901e-06,
+      "loss": 0.4757,
+      "step": 19065
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.481877264044309e-06,
+      "loss": 0.4377,
+      "step": 19066
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480877486569588e-06,
+      "loss": 0.4263,
+      "step": 19067
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479877765855303e-06,
+      "loss": 0.4292,
+      "step": 19068
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478878101914002e-06,
+      "loss": 0.417,
+      "step": 19069
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477878494758245e-06,
+      "loss": 0.4653,
+      "step": 19070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.476878944400592e-06,
+      "loss": 0.4542,
+      "step": 19071
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475879450853588e-06,
+      "loss": 0.3881,
+      "step": 19072
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474880014129793e-06,
+      "loss": 0.4118,
+      "step": 19073
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.473880634241762e-06,
+      "loss": 0.4429,
+      "step": 19074
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472881311202039e-06,
+      "loss": 0.4435,
+      "step": 19075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471882045023183e-06,
+      "loss": 0.3911,
+      "step": 19076
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.470882835717744e-06,
+      "loss": 0.46,
+      "step": 19077
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469883683298266e-06,
+      "loss": 0.4379,
+      "step": 19078
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468884587777302e-06,
+      "loss": 0.4769,
+      "step": 19079
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4678855491674065e-06,
+      "loss": 0.4935,
+      "step": 19080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4668865674811175e-06,
+      "loss": 0.3778,
+      "step": 19081
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.465887642730987e-06,
+      "loss": 0.4267,
+      "step": 19082
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464888774929566e-06,
+      "loss": 0.4692,
+      "step": 19083
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463889964089392e-06,
+      "loss": 0.4306,
+      "step": 19084
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4628912102230135e-06,
+      "loss": 0.4431,
+      "step": 19085
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461892513342981e-06,
+      "loss": 0.4209,
+      "step": 19086
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460893873461831e-06,
+      "loss": 0.4173,
+      "step": 19087
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.459895290592105e-06,
+      "loss": 0.4714,
+      "step": 19088
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458896764746352e-06,
+      "loss": 0.4305,
+      "step": 19089
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4578982959371066e-06,
+      "loss": 0.4353,
+      "step": 19090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.456899884176912e-06,
+      "loss": 0.4521,
+      "step": 19091
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455901529478315e-06,
+      "loss": 0.4683,
+      "step": 19092
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.454903231853843e-06,
+      "loss": 0.4273,
+      "step": 19093
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453904991316044e-06,
+      "loss": 0.4165,
+      "step": 19094
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452906807877454e-06,
+      "loss": 0.4737,
+      "step": 19095
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451908681550606e-06,
+      "loss": 0.4477,
+      "step": 19096
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450910612348044e-06,
+      "loss": 0.4553,
+      "step": 19097
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449912600282296e-06,
+      "loss": 0.3952,
+      "step": 19098
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448914645365904e-06,
+      "loss": 0.5091,
+      "step": 19099
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.447916747611395e-06,
+      "loss": 0.4765,
+      "step": 19100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446918907031307e-06,
+      "loss": 0.4551,
+      "step": 19101
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445921123638176e-06,
+      "loss": 0.4147,
+      "step": 19102
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.444923397444526e-06,
+      "loss": 0.4433,
+      "step": 19103
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443925728462894e-06,
+      "loss": 0.4328,
+      "step": 19104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442928116705815e-06,
+      "loss": 0.4307,
+      "step": 19105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.441930562185809e-06,
+      "loss": 0.4374,
+      "step": 19106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.440933064915414e-06,
+      "loss": 0.4182,
+      "step": 19107
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.439935624907151e-06,
+      "loss": 0.4501,
+      "step": 19108
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438938242173557e-06,
+      "loss": 0.4246,
+      "step": 19109
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437940916727149e-06,
+      "loss": 0.4439,
+      "step": 19110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4369436485804636e-06,
+      "loss": 0.4371,
+      "step": 19111
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435946437746018e-06,
+      "loss": 0.4786,
+      "step": 19112
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.434949284236339e-06,
+      "loss": 0.4245,
+      "step": 19113
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4339521880639575e-06,
+      "loss": 0.4546,
+      "step": 19114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432955149241389e-06,
+      "loss": 0.441,
+      "step": 19115
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4319581677811594e-06,
+      "loss": 0.393,
+      "step": 19116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.430961243695794e-06,
+      "loss": 0.4474,
+      "step": 19117
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429964376997808e-06,
+      "loss": 0.4116,
+      "step": 19118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.42896756769973e-06,
+      "loss": 0.4171,
+      "step": 19119
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.427970815814074e-06,
+      "loss": 0.4552,
+      "step": 19120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.426974121353357e-06,
+      "loss": 0.423,
+      "step": 19121
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425977484330102e-06,
+      "loss": 0.3855,
+      "step": 19122
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.424980904756829e-06,
+      "loss": 0.4761,
+      "step": 19123
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4239843826460485e-06,
+      "loss": 0.4299,
+      "step": 19124
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4229879180102804e-06,
+      "loss": 0.421,
+      "step": 19125
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.421991510862044e-06,
+      "loss": 0.4855,
+      "step": 19126
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.420995161213847e-06,
+      "loss": 0.4089,
+      "step": 19127
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419998869078207e-06,
+      "loss": 0.4869,
+      "step": 19128
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419002634467643e-06,
+      "loss": 0.3971,
+      "step": 19129
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418006457394661e-06,
+      "loss": 0.4149,
+      "step": 19130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.417010337871771e-06,
+      "loss": 0.4314,
+      "step": 19131
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.416014275911492e-06,
+      "loss": 0.4766,
+      "step": 19132
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415018271526326e-06,
+      "loss": 0.5041,
+      "step": 19133
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414022324728787e-06,
+      "loss": 0.4702,
+      "step": 19134
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413026435531389e-06,
+      "loss": 0.4536,
+      "step": 19135
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412030603946632e-06,
+      "loss": 0.4142,
+      "step": 19136
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4110348299870276e-06,
+      "loss": 0.421,
+      "step": 19137
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410039113665082e-06,
+      "loss": 0.422,
+      "step": 19138
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409043454993307e-06,
+      "loss": 0.416,
+      "step": 19139
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.408047853984203e-06,
+      "loss": 0.4566,
+      "step": 19140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407052310650272e-06,
+      "loss": 0.4644,
+      "step": 19141
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406056825004024e-06,
+      "loss": 0.4508,
+      "step": 19142
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4050613970579554e-06,
+      "loss": 0.445,
+      "step": 19143
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404066026824574e-06,
+      "loss": 0.4299,
+      "step": 19144
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.403070714316384e-06,
+      "loss": 0.3775,
+      "step": 19145
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402075459545879e-06,
+      "loss": 0.3951,
+      "step": 19146
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401080262525564e-06,
+      "loss": 0.4722,
+      "step": 19147
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400085123267944e-06,
+      "loss": 0.3951,
+      "step": 19148
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3990900417855065e-06,
+      "loss": 0.4384,
+      "step": 19149
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398095018090761e-06,
+      "loss": 0.4252,
+      "step": 19150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397100052196198e-06,
+      "loss": 0.4558,
+      "step": 19151
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396105144114314e-06,
+      "loss": 0.4566,
+      "step": 19152
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.395110293857606e-06,
+      "loss": 0.4181,
+      "step": 19153
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.394115501438576e-06,
+      "loss": 0.3665,
+      "step": 19154
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393120766869708e-06,
+      "loss": 0.4127,
+      "step": 19155
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392126090163503e-06,
+      "loss": 0.426,
+      "step": 19156
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391131471332455e-06,
+      "loss": 0.4069,
+      "step": 19157
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3901369103890496e-06,
+      "loss": 0.4016,
+      "step": 19158
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389142407345783e-06,
+      "loss": 0.4431,
+      "step": 19159
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.38814796221515e-06,
+      "loss": 0.4691,
+      "step": 19160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387153575009637e-06,
+      "loss": 0.3928,
+      "step": 19161
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386159245741729e-06,
+      "loss": 0.415,
+      "step": 19162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.385164974423925e-06,
+      "loss": 0.4302,
+      "step": 19163
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384170761068701e-06,
+      "loss": 0.5179,
+      "step": 19164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3831766056885515e-06,
+      "loss": 0.4714,
+      "step": 19165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382182508295968e-06,
+      "loss": 0.4695,
+      "step": 19166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3811884689034245e-06,
+      "loss": 0.4126,
+      "step": 19167
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380194487523414e-06,
+      "loss": 0.4368,
+      "step": 19168
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.379200564168423e-06,
+      "loss": 0.3948,
+      "step": 19169
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3782066988509276e-06,
+      "loss": 0.4498,
+      "step": 19170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.377212891583419e-06,
+      "loss": 0.4773,
+      "step": 19171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.376219142378372e-06,
+      "loss": 0.4195,
+      "step": 19172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3752254512482746e-06,
+      "loss": 0.4731,
+      "step": 19173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374231818205602e-06,
+      "loss": 0.4429,
+      "step": 19174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.373238243262835e-06,
+      "loss": 0.4358,
+      "step": 19175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372244726432461e-06,
+      "loss": 0.3782,
+      "step": 19176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.371251267726946e-06,
+      "loss": 0.3815,
+      "step": 19177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3702578671587744e-06,
+      "loss": 0.4627,
+      "step": 19178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36926452474043e-06,
+      "loss": 0.4096,
+      "step": 19179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368271240484375e-06,
+      "loss": 0.4212,
+      "step": 19180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367278014403099e-06,
+      "loss": 0.4515,
+      "step": 19181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366284846509067e-06,
+      "loss": 0.4764,
+      "step": 19182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365291736814759e-06,
+      "loss": 0.4654,
+      "step": 19183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.364298685332643e-06,
+      "loss": 0.4359,
+      "step": 19184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3633056920751994e-06,
+      "loss": 0.4838,
+      "step": 19185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.362312757054892e-06,
+      "loss": 0.4653,
+      "step": 19186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361319880284196e-06,
+      "loss": 0.4279,
+      "step": 19187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360327061775586e-06,
+      "loss": 0.4411,
+      "step": 19188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.359334301541524e-06,
+      "loss": 0.4146,
+      "step": 19189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358341599594483e-06,
+      "loss": 0.4489,
+      "step": 19190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357348955946935e-06,
+      "loss": 0.4266,
+      "step": 19191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.356356370611342e-06,
+      "loss": 0.4314,
+      "step": 19192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.355363843600175e-06,
+      "loss": 0.441,
+      "step": 19193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3543713749259e-06,
+      "loss": 0.4626,
+      "step": 19194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353378964600975e-06,
+      "loss": 0.4296,
+      "step": 19195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3523866126378725e-06,
+      "loss": 0.3996,
+      "step": 19196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3513943190490554e-06,
+      "loss": 0.4657,
+      "step": 19197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3504020838469835e-06,
+      "loss": 0.442,
+      "step": 19198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.349409907044123e-06,
+      "loss": 0.4315,
+      "step": 19199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348417788652938e-06,
+      "loss": 0.4277,
+      "step": 19200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347425728685881e-06,
+      "loss": 0.4518,
+      "step": 19201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346433727155422e-06,
+      "loss": 0.444,
+      "step": 19202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3454417840740125e-06,
+      "loss": 0.3909,
+      "step": 19203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344449899454118e-06,
+      "loss": 0.4179,
+      "step": 19204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34345807330819e-06,
+      "loss": 0.4396,
+      "step": 19205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3424663056486936e-06,
+      "loss": 0.3925,
+      "step": 19206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3414745964880775e-06,
+      "loss": 0.4321,
+      "step": 19207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.340482945838801e-06,
+      "loss": 0.4124,
+      "step": 19208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33949135371332e-06,
+      "loss": 0.4344,
+      "step": 19209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3384998201240924e-06,
+      "loss": 0.4328,
+      "step": 19210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337508345083564e-06,
+      "loss": 0.4893,
+      "step": 19211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3365169286041954e-06,
+      "loss": 0.4083,
+      "step": 19212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335525570698432e-06,
+      "loss": 0.4143,
+      "step": 19213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.334534271378732e-06,
+      "loss": 0.4353,
+      "step": 19214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33354303065754e-06,
+      "loss": 0.439,
+      "step": 19215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.332551848547311e-06,
+      "loss": 0.4258,
+      "step": 19216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33156072506049e-06,
+      "loss": 0.4392,
+      "step": 19217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.330569660209528e-06,
+      "loss": 0.4084,
+      "step": 19218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.329578654006875e-06,
+      "loss": 0.3918,
+      "step": 19219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328587706464973e-06,
+      "loss": 0.4504,
+      "step": 19220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32759681759627e-06,
+      "loss": 0.3799,
+      "step": 19221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326605987413217e-06,
+      "loss": 0.4359,
+      "step": 19222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32561521592825e-06,
+      "loss": 0.4662,
+      "step": 19223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.324624503153823e-06,
+      "loss": 0.4429,
+      "step": 19224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3236338491023745e-06,
+      "loss": 0.4739,
+      "step": 19225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322643253786341e-06,
+      "loss": 0.3649,
+      "step": 19226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321652717218171e-06,
+      "loss": 0.4376,
+      "step": 19227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.320662239410308e-06,
+      "loss": 0.4704,
+      "step": 19228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319671820375186e-06,
+      "loss": 0.414,
+      "step": 19229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318681460125249e-06,
+      "loss": 0.4271,
+      "step": 19230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3176911586729374e-06,
+      "loss": 0.4416,
+      "step": 19231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3167009160306836e-06,
+      "loss": 0.3889,
+      "step": 19232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.31571073221093e-06,
+      "loss": 0.4312,
+      "step": 19233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.314720607226115e-06,
+      "loss": 0.4528,
+      "step": 19234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313730541088671e-06,
+      "loss": 0.4454,
+      "step": 19235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.31274053381103e-06,
+      "loss": 0.4175,
+      "step": 19236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311750585405635e-06,
+      "loss": 0.4022,
+      "step": 19237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310760695884911e-06,
+      "loss": 0.44,
+      "step": 19238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3097708652612945e-06,
+      "loss": 0.4474,
+      "step": 19239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308781093547224e-06,
+      "loss": 0.4462,
+      "step": 19240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3077913807551204e-06,
+      "loss": 0.4478,
+      "step": 19241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3068017268974215e-06,
+      "loss": 0.4378,
+      "step": 19242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305812131986559e-06,
+      "loss": 0.4143,
+      "step": 19243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304822596034954e-06,
+      "loss": 0.4237,
+      "step": 19244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.303833119055045e-06,
+      "loss": 0.4198,
+      "step": 19245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30284370105925e-06,
+      "loss": 0.508,
+      "step": 19246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301854342060007e-06,
+      "loss": 0.4235,
+      "step": 19247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3008650420697315e-06,
+      "loss": 0.4089,
+      "step": 19248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.299875801100856e-06,
+      "loss": 0.4622,
+      "step": 19249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298886619165805e-06,
+      "loss": 0.4754,
+      "step": 19250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.297897496276998e-06,
+      "loss": 0.4458,
+      "step": 19251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296908432446863e-06,
+      "loss": 0.4296,
+      "step": 19252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295919427687825e-06,
+      "loss": 0.4577,
+      "step": 19253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294930482012298e-06,
+      "loss": 0.425,
+      "step": 19254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293941595432712e-06,
+      "loss": 0.4267,
+      "step": 19255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2929527679614835e-06,
+      "loss": 0.4557,
+      "step": 19256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.291963999611027e-06,
+      "loss": 0.4652,
+      "step": 19257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290975290393768e-06,
+      "loss": 0.4655,
+      "step": 19258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.289986640322124e-06,
+      "loss": 0.461,
+      "step": 19259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.28899804940851e-06,
+      "loss": 0.5128,
+      "step": 19260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288009517665345e-06,
+      "loss": 0.4081,
+      "step": 19261
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287021045105045e-06,
+      "loss": 0.4553,
+      "step": 19262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.4136,
+      "step": 19263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285044277582696e-06,
+      "loss": 0.4469,
+      "step": 19264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2840559826454795e-06,
+      "loss": 0.3857,
+      "step": 19265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283067746940784e-06,
+      "loss": 0.3886,
+      "step": 19266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.282079570481018e-06,
+      "loss": 0.3457,
+      "step": 19267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.281091453278601e-06,
+      "loss": 0.402,
+      "step": 19268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280103395345936e-06,
+      "loss": 0.3459,
+      "step": 19269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.279115396695436e-06,
+      "loss": 0.3601,
+      "step": 19270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278127457339515e-06,
+      "loss": 0.3437,
+      "step": 19271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277139577290574e-06,
+      "loss": 0.3628,
+      "step": 19272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.276151756561026e-06,
+      "loss": 0.3574,
+      "step": 19273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.275163995163277e-06,
+      "loss": 0.3352,
+      "step": 19274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274176293109732e-06,
+      "loss": 0.3308,
+      "step": 19275
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2731886504128e-06,
+      "loss": 0.3661,
+      "step": 19276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2722010670848785e-06,
+      "loss": 0.3391,
+      "step": 19277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271213543138381e-06,
+      "loss": 0.3325,
+      "step": 19278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.270226078585703e-06,
+      "loss": 0.354,
+      "step": 19279
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.269238673439254e-06,
+      "loss": 0.3557,
+      "step": 19280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.268251327711427e-06,
+      "loss": 0.3349,
+      "step": 19281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.26726404141463e-06,
+      "loss": 0.3351,
+      "step": 19282
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.266276814561261e-06,
+      "loss": 0.3419,
+      "step": 19283
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265289647163725e-06,
+      "loss": 0.3386,
+      "step": 19284
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.264302539234412e-06,
+      "loss": 0.3309,
+      "step": 19285
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263315490785727e-06,
+      "loss": 0.3475,
+      "step": 19286
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.262328501830062e-06,
+      "loss": 0.3202,
+      "step": 19287
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261341572379819e-06,
+      "loss": 0.3334,
+      "step": 19288
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260354702447389e-06,
+      "loss": 0.3314,
+      "step": 19289
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.259367892045173e-06,
+      "loss": 0.3468,
+      "step": 19290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.258381141185558e-06,
+      "loss": 0.2863,
+      "step": 19291
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257394449880939e-06,
+      "loss": 0.3132,
+      "step": 19292
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.256407818143718e-06,
+      "loss": 0.3521,
+      "step": 19293
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255421245986276e-06,
+      "loss": 0.3299,
+      "step": 19294
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.254434733421008e-06,
+      "loss": 0.3215,
+      "step": 19295
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25344828046031e-06,
+      "loss": 0.3104,
+      "step": 19296
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252461887116564e-06,
+      "loss": 0.3179,
+      "step": 19297
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2514755534021655e-06,
+      "loss": 0.3384,
+      "step": 19298
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250489279329501e-06,
+      "loss": 0.3375,
+      "step": 19299
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249503064910952e-06,
+      "loss": 0.3292,
+      "step": 19300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.248516910158911e-06,
+      "loss": 0.3456,
+      "step": 19301
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.247530815085769e-06,
+      "loss": 0.2968,
+      "step": 19302
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246544779703902e-06,
+      "loss": 0.3747,
+      "step": 19303
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.245558804025697e-06,
+      "loss": 0.347,
+      "step": 19304
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.244572888063545e-06,
+      "loss": 0.3096,
+      "step": 19305
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.24358703182982e-06,
+      "loss": 0.3336,
+      "step": 19306
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.242601235336907e-06,
+      "loss": 0.3413,
+      "step": 19307
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.241615498597195e-06,
+      "loss": 0.3645,
+      "step": 19308
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240629821623055e-06,
+      "loss": 0.3286,
+      "step": 19309
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.23964420442687e-06,
+      "loss": 0.2854,
+      "step": 19310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.238658647021024e-06,
+      "loss": 0.3294,
+      "step": 19311
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237673149417889e-06,
+      "loss": 0.326,
+      "step": 19312
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2366877116298446e-06,
+      "loss": 0.3474,
+      "step": 19313
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.235702333669273e-06,
+      "loss": 0.3488,
+      "step": 19314
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.234717015548545e-06,
+      "loss": 0.3353,
+      "step": 19315
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.233731757280036e-06,
+      "loss": 0.3341,
+      "step": 19316
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.232746558876127e-06,
+      "loss": 0.3017,
+      "step": 19317
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231761420349186e-06,
+      "loss": 0.3323,
+      "step": 19318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.230776341711592e-06,
+      "loss": 0.3125,
+      "step": 19319
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22979132297571e-06,
+      "loss": 0.3292,
+      "step": 19320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22880636415392e-06,
+      "loss": 0.3691,
+      "step": 19321
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227821465258584e-06,
+      "loss": 0.3134,
+      "step": 19322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.226836626302081e-06,
+      "loss": 0.2747,
+      "step": 19323
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225851847296779e-06,
+      "loss": 0.3267,
+      "step": 19324
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2248671282550415e-06,
+      "loss": 0.2799,
+      "step": 19325
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22388246918924e-06,
+      "loss": 0.2983,
+      "step": 19326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222897870111746e-06,
+      "loss": 0.3539,
+      "step": 19327
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.221913331034919e-06,
+      "loss": 0.3675,
+      "step": 19328
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220928851971132e-06,
+      "loss": 0.3228,
+      "step": 19329
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219944432932747e-06,
+      "loss": 0.3492,
+      "step": 19330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218960073932122e-06,
+      "loss": 0.3459,
+      "step": 19331
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.217975774981628e-06,
+      "loss": 0.3689,
+      "step": 19332
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2169915360936285e-06,
+      "loss": 0.3336,
+      "step": 19333
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216007357280479e-06,
+      "loss": 0.3383,
+      "step": 19334
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.215023238554546e-06,
+      "loss": 0.303,
+      "step": 19335
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214039179928194e-06,
+      "loss": 0.3537,
+      "step": 19336
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213055181413773e-06,
+      "loss": 0.3212,
+      "step": 19337
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212071243023646e-06,
+      "loss": 0.317,
+      "step": 19338
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.211087364770178e-06,
+      "loss": 0.3525,
+      "step": 19339
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.21010354666572e-06,
+      "loss": 0.3145,
+      "step": 19340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209119788722625e-06,
+      "loss": 0.315,
+      "step": 19341
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20813609095326e-06,
+      "loss": 0.3346,
+      "step": 19342
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.207152453369968e-06,
+      "loss": 0.3405,
+      "step": 19343
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206168875985111e-06,
+      "loss": 0.309,
+      "step": 19344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205185358811044e-06,
+      "loss": 0.3425,
+      "step": 19345
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2042019018601155e-06,
+      "loss": 0.3085,
+      "step": 19346
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.203218505144678e-06,
+      "loss": 0.3622,
+      "step": 19347
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202235168677089e-06,
+      "loss": 0.3537,
+      "step": 19348
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201251892469692e-06,
+      "loss": 0.318,
+      "step": 19349
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200268676534843e-06,
+      "loss": 0.3395,
+      "step": 19350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.199285520884887e-06,
+      "loss": 0.3068,
+      "step": 19351
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.198302425532171e-06,
+      "loss": 0.2934,
+      "step": 19352
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197319390489046e-06,
+      "loss": 0.3091,
+      "step": 19353
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196336415767862e-06,
+      "loss": 0.3082,
+      "step": 19354
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.195353501380957e-06,
+      "loss": 0.3062,
+      "step": 19355
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194370647340683e-06,
+      "loss": 0.3465,
+      "step": 19356
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193387853659382e-06,
+      "loss": 0.3163,
+      "step": 19357
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.192405120349402e-06,
+      "loss": 0.2891,
+      "step": 19358
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191422447423079e-06,
+      "loss": 0.3286,
+      "step": 19359
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.190439834892764e-06,
+      "loss": 0.3371,
+      "step": 19360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189457282770793e-06,
+      "loss": 0.3592,
+      "step": 19361
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.188474791069504e-06,
+      "loss": 0.3408,
+      "step": 19362
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.187492359801243e-06,
+      "loss": 0.2768,
+      "step": 19363
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18650998897835e-06,
+      "loss": 0.3418,
+      "step": 19364
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185527678613158e-06,
+      "loss": 0.2987,
+      "step": 19365
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.184545428718006e-06,
+      "loss": 0.3191,
+      "step": 19366
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.183563239305239e-06,
+      "loss": 0.3334,
+      "step": 19367
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1825811103871834e-06,
+      "loss": 0.3378,
+      "step": 19368
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181599041976177e-06,
+      "loss": 0.3187,
+      "step": 19369
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180617034084563e-06,
+      "loss": 0.3436,
+      "step": 19370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179635086724668e-06,
+      "loss": 0.3481,
+      "step": 19371
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178653199908822e-06,
+      "loss": 0.3061,
+      "step": 19372
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.177671373649367e-06,
+      "loss": 0.3183,
+      "step": 19373
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.176689607958624e-06,
+      "loss": 0.3153,
+      "step": 19374
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.17570790284893e-06,
+      "loss": 0.3497,
+      "step": 19375
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.174726258332619e-06,
+      "loss": 0.3687,
+      "step": 19376
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.173744674422012e-06,
+      "loss": 0.3221,
+      "step": 19377
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172763151129443e-06,
+      "loss": 0.3395,
+      "step": 19378
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1717816884672435e-06,
+      "loss": 0.3514,
+      "step": 19379
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170800286447732e-06,
+      "loss": 0.3607,
+      "step": 19380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169818945083243e-06,
+      "loss": 0.3548,
+      "step": 19381
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.168837664386095e-06,
+      "loss": 0.3108,
+      "step": 19382
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1678564443686195e-06,
+      "loss": 0.3439,
+      "step": 19383
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166875285043136e-06,
+      "loss": 0.2954,
+      "step": 19384
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.165894186421971e-06,
+      "loss": 0.3192,
+      "step": 19385
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.164913148517443e-06,
+      "loss": 0.3407,
+      "step": 19386
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163932171341877e-06,
+      "loss": 0.33,
+      "step": 19387
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1629512549075974e-06,
+      "loss": 0.3429,
+      "step": 19388
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1619703992269165e-06,
+      "loss": 0.325,
+      "step": 19389
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.160989604312159e-06,
+      "loss": 0.3242,
+      "step": 19390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.160008870175647e-06,
+      "loss": 0.3228,
+      "step": 19391
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.159028196829689e-06,
+      "loss": 0.2947,
+      "step": 19392
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158047584286614e-06,
+      "loss": 0.335,
+      "step": 19393
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.157067032558729e-06,
+      "loss": 0.3303,
+      "step": 19394
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.156086541658356e-06,
+      "loss": 0.3138,
+      "step": 19395
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155106111597803e-06,
+      "loss": 0.3129,
+      "step": 19396
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1541257423893894e-06,
+      "loss": 0.3204,
+      "step": 19397
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.153145434045432e-06,
+      "loss": 0.349,
+      "step": 19398
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.152165186578235e-06,
+      "loss": 0.3085,
+      "step": 19399
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.151185000000114e-06,
+      "loss": 0.3177,
+      "step": 19400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150204874323387e-06,
+      "loss": 0.3537,
+      "step": 19401
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149224809560352e-06,
+      "loss": 0.348,
+      "step": 19402
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1482448057233305e-06,
+      "loss": 0.3366,
+      "step": 19403
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.147264862824626e-06,
+      "loss": 0.2781,
+      "step": 19404
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146284980876542e-06,
+      "loss": 0.329,
+      "step": 19405
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1453051598913915e-06,
+      "loss": 0.3232,
+      "step": 19406
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144325399881482e-06,
+      "loss": 0.3363,
+      "step": 19407
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143345700859116e-06,
+      "loss": 0.3501,
+      "step": 19408
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.142366062836599e-06,
+      "loss": 0.343,
+      "step": 19409
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1413864858262396e-06,
+      "loss": 0.361,
+      "step": 19410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.140406969840335e-06,
+      "loss": 0.3459,
+      "step": 19411
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.139427514891191e-06,
+      "loss": 0.3378,
+      "step": 19412
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138448120991113e-06,
+      "loss": 0.3536,
+      "step": 19413
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.137468788152398e-06,
+      "loss": 0.313,
+      "step": 19414
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136489516387345e-06,
+      "loss": 0.3294,
+      "step": 19415
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13551030570826e-06,
+      "loss": 0.3206,
+      "step": 19416
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.134531156127433e-06,
+      "loss": 0.2872,
+      "step": 19417
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.133552067657168e-06,
+      "loss": 0.286,
+      "step": 19418
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132573040309765e-06,
+      "loss": 0.3415,
+      "step": 19419
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131594074097513e-06,
+      "loss": 0.2953,
+      "step": 19420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130615169032713e-06,
+      "loss": 0.2918,
+      "step": 19421
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.129636325127663e-06,
+      "loss": 0.3292,
+      "step": 19422
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.12865754239465e-06,
+      "loss": 0.3407,
+      "step": 19423
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127678820845974e-06,
+      "loss": 0.2998,
+      "step": 19424
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1267001604939245e-06,
+      "loss": 0.3219,
+      "step": 19425
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1257215613507916e-06,
+      "loss": 0.3193,
+      "step": 19426
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.124743023428867e-06,
+      "loss": 0.3063,
+      "step": 19427
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123764546740444e-06,
+      "loss": 0.3217,
+      "step": 19428
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.122786131297817e-06,
+      "loss": 0.3099,
+      "step": 19429
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121807777113263e-06,
+      "loss": 0.3083,
+      "step": 19430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120829484199077e-06,
+      "loss": 0.3503,
+      "step": 19431
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.119851252567551e-06,
+      "loss": 0.2903,
+      "step": 19432
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.118873082230962e-06,
+      "loss": 0.3786,
+      "step": 19433
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117894973201604e-06,
+      "loss": 0.2604,
+      "step": 19434
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.116916925491758e-06,
+      "loss": 0.3291,
+      "step": 19435
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115938939113707e-06,
+      "loss": 0.3294,
+      "step": 19436
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114961014079734e-06,
+      "loss": 0.3194,
+      "step": 19437
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113983150402131e-06,
+      "loss": 0.3392,
+      "step": 19438
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113005348093167e-06,
+      "loss": 0.3153,
+      "step": 19439
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112027607165131e-06,
+      "loss": 0.3319,
+      "step": 19440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1110499276303045e-06,
+      "loss": 0.3479,
+      "step": 19441
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110072309500962e-06,
+      "loss": 0.3587,
+      "step": 19442
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109094752789384e-06,
+      "loss": 0.3057,
+      "step": 19443
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108117257507854e-06,
+      "loss": 0.3428,
+      "step": 19444
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.107139823668647e-06,
+      "loss": 0.2982,
+      "step": 19445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106162451284031e-06,
+      "loss": 0.3157,
+      "step": 19446
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.105185140366296e-06,
+      "loss": 0.2985,
+      "step": 19447
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.104207890927704e-06,
+      "loss": 0.3048,
+      "step": 19448
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103230702980535e-06,
+      "loss": 0.3415,
+      "step": 19449
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102253576537066e-06,
+      "loss": 0.3462,
+      "step": 19450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.101276511609563e-06,
+      "loss": 0.3121,
+      "step": 19451
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100299508210303e-06,
+      "loss": 0.2899,
+      "step": 19452
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099322566351556e-06,
+      "loss": 0.3525,
+      "step": 19453
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0983456860455914e-06,
+      "loss": 0.3297,
+      "step": 19454
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0973688673046815e-06,
+      "loss": 0.3694,
+      "step": 19455
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.096392110141091e-06,
+      "loss": 0.3344,
+      "step": 19456
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.095415414567093e-06,
+      "loss": 0.2887,
+      "step": 19457
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094438780594948e-06,
+      "loss": 0.3181,
+      "step": 19458
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.093462208236931e-06,
+      "loss": 0.314,
+      "step": 19459
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092485697505299e-06,
+      "loss": 0.3431,
+      "step": 19460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.091509248412323e-06,
+      "loss": 0.3352,
+      "step": 19461
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090532860970268e-06,
+      "loss": 0.3362,
+      "step": 19462
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.089556535191391e-06,
+      "loss": 0.2886,
+      "step": 19463
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088580271087961e-06,
+      "loss": 0.3338,
+      "step": 19464
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08760406867224e-06,
+      "loss": 0.3275,
+      "step": 19465
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086627927956488e-06,
+      "loss": 0.3438,
+      "step": 19466
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08565184895296e-06,
+      "loss": 0.2993,
+      "step": 19467
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08467583167392e-06,
+      "loss": 0.2928,
+      "step": 19468
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0836998761316316e-06,
+      "loss": 0.3139,
+      "step": 19469
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.082723982338343e-06,
+      "loss": 0.314,
+      "step": 19470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081748150306317e-06,
+      "loss": 0.3401,
+      "step": 19471
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080772380047814e-06,
+      "loss": 0.3424,
+      "step": 19472
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0797966715750815e-06,
+      "loss": 0.3357,
+      "step": 19473
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.078821024900378e-06,
+      "loss": 0.271,
+      "step": 19474
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077845440035961e-06,
+      "loss": 0.3048,
+      "step": 19475
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076869916994083e-06,
+      "loss": 0.3101,
+      "step": 19476
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.075894455786989e-06,
+      "loss": 0.3267,
+      "step": 19477
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074919056426941e-06,
+      "loss": 0.3164,
+      "step": 19478
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073943718926182e-06,
+      "loss": 0.3,
+      "step": 19479
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.072968443296964e-06,
+      "loss": 0.3321,
+      "step": 19480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071993229551544e-06,
+      "loss": 0.3159,
+      "step": 19481
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071018077702161e-06,
+      "loss": 0.3058,
+      "step": 19482
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070042987761066e-06,
+      "loss": 0.3386,
+      "step": 19483
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0690679597405124e-06,
+      "loss": 0.3178,
+      "step": 19484
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0680929936527356e-06,
+      "loss": 0.3143,
+      "step": 19485
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.067118089509993e-06,
+      "loss": 0.3424,
+      "step": 19486
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.066143247324517e-06,
+      "loss": 0.3515,
+      "step": 19487
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065168467108565e-06,
+      "loss": 0.2787,
+      "step": 19488
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.064193748874369e-06,
+      "loss": 0.3035,
+      "step": 19489
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063219092634178e-06,
+      "loss": 0.3055,
+      "step": 19490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.062244498400228e-06,
+      "loss": 0.3463,
+      "step": 19491
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061269966184766e-06,
+      "loss": 0.3298,
+      "step": 19492
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.060295496000032e-06,
+      "loss": 0.3116,
+      "step": 19493
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05932108785826e-06,
+      "loss": 0.3145,
+      "step": 19494
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0583467417716914e-06,
+      "loss": 0.3313,
+      "step": 19495
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057372457752567e-06,
+      "loss": 0.3404,
+      "step": 19496
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05639823581312e-06,
+      "loss": 0.336,
+      "step": 19497
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0554240759655906e-06,
+      "loss": 0.3112,
+      "step": 19498
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0544499782222115e-06,
+      "loss": 0.3127,
+      "step": 19499
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053475942595214e-06,
+      "loss": 0.3504,
+      "step": 19500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.052501969096837e-06,
+      "loss": 0.3314,
+      "step": 19501
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05152805773931e-06,
+      "loss": 0.3318,
+      "step": 19502
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.050554208534874e-06,
+      "loss": 0.3455,
+      "step": 19503
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.049580421495749e-06,
+      "loss": 0.3385,
+      "step": 19504
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048606696634171e-06,
+      "loss": 0.3671,
+      "step": 19505
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.047633033962375e-06,
+      "loss": 0.2781,
+      "step": 19506
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.04665943349258e-06,
+      "loss": 0.3426,
+      "step": 19507
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.045685895237026e-06,
+      "loss": 0.3018,
+      "step": 19508
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0447124192079335e-06,
+      "loss": 0.3208,
+      "step": 19509
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0437390054175275e-06,
+      "loss": 0.3491,
+      "step": 19510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.042765653878035e-06,
+      "loss": 0.3148,
+      "step": 19511
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.041792364601689e-06,
+      "loss": 0.3402,
+      "step": 19512
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040819137600705e-06,
+      "loss": 0.3221,
+      "step": 19513
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.039845972887309e-06,
+      "loss": 0.3507,
+      "step": 19514
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03887287047373e-06,
+      "loss": 0.3508,
+      "step": 19515
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.037899830372183e-06,
+      "loss": 0.3261,
+      "step": 19516
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03692685259489e-06,
+      "loss": 0.316,
+      "step": 19517
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0359539371540785e-06,
+      "loss": 0.331,
+      "step": 19518
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034981084061964e-06,
+      "loss": 0.3288,
+      "step": 19519
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034008293330761e-06,
+      "loss": 0.3204,
+      "step": 19520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.033035564972695e-06,
+      "loss": 0.3308,
+      "step": 19521
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.032062898999978e-06,
+      "loss": 0.3495,
+      "step": 19522
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.031090295424828e-06,
+      "loss": 0.3432,
+      "step": 19523
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.030117754259467e-06,
+      "loss": 0.3201,
+      "step": 19524
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0291452755161e-06,
+      "loss": 0.3262,
+      "step": 19525
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.028172859206948e-06,
+      "loss": 0.3006,
+      "step": 19526
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.027200505344228e-06,
+      "loss": 0.313,
+      "step": 19527
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.026228213940143e-06,
+      "loss": 0.3274,
+      "step": 19528
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025255985006914e-06,
+      "loss": 0.3233,
+      "step": 19529
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024283818556749e-06,
+      "loss": 0.3282,
+      "step": 19530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.023311714601854e-06,
+      "loss": 0.3287,
+      "step": 19531
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022339673154442e-06,
+      "loss": 0.3908,
+      "step": 19532
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021367694226727e-06,
+      "loss": 0.3116,
+      "step": 19533
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0203957778309085e-06,
+      "loss": 0.3247,
+      "step": 19534
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019423923979198e-06,
+      "loss": 0.3603,
+      "step": 19535
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.018452132683806e-06,
+      "loss": 0.3143,
+      "step": 19536
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0174804039569295e-06,
+      "loss": 0.332,
+      "step": 19537
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016508737810779e-06,
+      "loss": 0.3379,
+      "step": 19538
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015537134257562e-06,
+      "loss": 0.3341,
+      "step": 19539
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.014565593309477e-06,
+      "loss": 0.2927,
+      "step": 19540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013594114978724e-06,
+      "loss": 0.3244,
+      "step": 19541
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012622699277507e-06,
+      "loss": 0.293,
+      "step": 19542
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011651346218034e-06,
+      "loss": 0.3207,
+      "step": 19543
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010680055812494e-06,
+      "loss": 0.3492,
+      "step": 19544
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.009708828073092e-06,
+      "loss": 0.3568,
+      "step": 19545
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008737663012031e-06,
+      "loss": 0.321,
+      "step": 19546
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007766560641502e-06,
+      "loss": 0.3152,
+      "step": 19547
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0067955209737016e-06,
+      "loss": 0.3397,
+      "step": 19548
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005824544020837e-06,
+      "loss": 0.3361,
+      "step": 19549
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0048536297950924e-06,
+      "loss": 0.3399,
+      "step": 19550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.003882778308663e-06,
+      "loss": 0.3337,
+      "step": 19551
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.002911989573749e-06,
+      "loss": 0.3477,
+      "step": 19552
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0019412636025375e-06,
+      "loss": 0.3157,
+      "step": 19553
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000970600407223e-06,
+      "loss": 0.3123,
+      "step": 19554
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.3068,
+      "step": 19555
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.999029462393058e-06,
+      "loss": 0.3411,
+      "step": 19556
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.998058987598583e-06,
+      "loss": 0.2983,
+      "step": 19557
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.997088575628771e-06,
+      "loss": 0.3602,
+      "step": 19558
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996118226495803e-06,
+      "loss": 0.3439,
+      "step": 19559
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995147940211874e-06,
+      "loss": 0.3045,
+      "step": 19560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994177716789163e-06,
+      "loss": 0.359,
+      "step": 19561
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993207556239865e-06,
+      "loss": 0.2899,
+      "step": 19562
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9922374585761564e-06,
+      "loss": 0.3539,
+      "step": 19563
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.99126742381023e-06,
+      "loss": 0.3476,
+      "step": 19564
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.99029745195426e-06,
+      "loss": 0.3222,
+      "step": 19565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.989327543020436e-06,
+      "loss": 0.3601,
+      "step": 19566
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988357697020943e-06,
+      "loss": 0.3301,
+      "step": 19567
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987387913967952e-06,
+      "loss": 0.3346,
+      "step": 19568
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9864181938736514e-06,
+      "loss": 0.3078,
+      "step": 19569
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985448536750223e-06,
+      "loss": 0.3314,
+      "step": 19570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984478942609836e-06,
+      "loss": 0.3226,
+      "step": 19571
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.98350941146468e-06,
+      "loss": 0.3204,
+      "step": 19572
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982539943326926e-06,
+      "loss": 0.3382,
+      "step": 19573
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.981570538208747e-06,
+      "loss": 0.3202,
+      "step": 19574
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.980601196122324e-06,
+      "loss": 0.3146,
+      "step": 19575
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.979631917079831e-06,
+      "loss": 0.3576,
+      "step": 19576
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978662701093446e-06,
+      "loss": 0.3377,
+      "step": 19577
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9776935481753355e-06,
+      "loss": 0.3473,
+      "step": 19578
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.976724458337675e-06,
+      "loss": 0.3424,
+      "step": 19579
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975755431592641e-06,
+      "loss": 0.3437,
+      "step": 19580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974786467952395e-06,
+      "loss": 0.3246,
+      "step": 19581
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973817567429117e-06,
+      "loss": 0.3351,
+      "step": 19582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.972848730034972e-06,
+      "loss": 0.3486,
+      "step": 19583
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971879955782123e-06,
+      "loss": 0.3482,
+      "step": 19584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.970911244682745e-06,
+      "loss": 0.3306,
+      "step": 19585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.969942596749007e-06,
+      "loss": 0.3257,
+      "step": 19586
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968974011993067e-06,
+      "loss": 0.3003,
+      "step": 19587
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968005490427097e-06,
+      "loss": 0.3085,
+      "step": 19588
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967037032063262e-06,
+      "loss": 0.2785,
+      "step": 19589
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9660686369137215e-06,
+      "loss": 0.3404,
+      "step": 19590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.96510030499064e-06,
+      "loss": 0.3147,
+      "step": 19591
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964132036306184e-06,
+      "loss": 0.3454,
+      "step": 19592
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.963163830872511e-06,
+      "loss": 0.3244,
+      "step": 19593
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9621956887017805e-06,
+      "loss": 0.3567,
+      "step": 19594
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961227609806157e-06,
+      "loss": 0.3309,
+      "step": 19595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9602595941977936e-06,
+      "loss": 0.3322,
+      "step": 19596
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9592916418888516e-06,
+      "loss": 0.3216,
+      "step": 19597
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958323752891493e-06,
+      "loss": 0.3474,
+      "step": 19598
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957355927217866e-06,
+      "loss": 0.3191,
+      "step": 19599
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956388164880131e-06,
+      "loss": 0.3575,
+      "step": 19600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9554204658904485e-06,
+      "loss": 0.3305,
+      "step": 19601
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9544528302609615e-06,
+      "loss": 0.2989,
+      "step": 19602
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.953485258003835e-06,
+      "loss": 0.291,
+      "step": 19603
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.952517749131216e-06,
+      "loss": 0.306,
+      "step": 19604
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.951550303655253e-06,
+      "loss": 0.3031,
+      "step": 19605
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9505829215881e-06,
+      "loss": 0.2994,
+      "step": 19606
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949615602941913e-06,
+      "loss": 0.3081,
+      "step": 19607
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9486483477288325e-06,
+      "loss": 0.3236,
+      "step": 19608
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947681155961012e-06,
+      "loss": 0.2999,
+      "step": 19609
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.946714027650599e-06,
+      "loss": 0.3216,
+      "step": 19610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.945746962809745e-06,
+      "loss": 0.3381,
+      "step": 19611
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.944779961450588e-06,
+      "loss": 0.3147,
+      "step": 19612
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943813023585281e-06,
+      "loss": 0.3382,
+      "step": 19613
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942846149225966e-06,
+      "loss": 0.3163,
+      "step": 19614
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941879338384783e-06,
+      "loss": 0.3003,
+      "step": 19615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.94091259107388e-06,
+      "loss": 0.324,
+      "step": 19616
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9399459073054e-06,
+      "loss": 0.3091,
+      "step": 19617
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.93897928709148e-06,
+      "loss": 0.3211,
+      "step": 19618
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938012730444264e-06,
+      "loss": 0.3397,
+      "step": 19619
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.937046237375894e-06,
+      "loss": 0.3229,
+      "step": 19620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936079807898504e-06,
+      "loss": 0.3277,
+      "step": 19621
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935113442024235e-06,
+      "loss": 0.3374,
+      "step": 19622
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.93414713976523e-06,
+      "loss": 0.2741,
+      "step": 19623
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933180901133618e-06,
+      "loss": 0.2839,
+      "step": 19624
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932214726141535e-06,
+      "loss": 0.3305,
+      "step": 19625
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9312486148011245e-06,
+      "loss": 0.3176,
+      "step": 19626
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930282567124509e-06,
+      "loss": 0.295,
+      "step": 19627
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9293165831238295e-06,
+      "loss": 0.3212,
+      "step": 19628
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.928350662811221e-06,
+      "loss": 0.3508,
+      "step": 19629
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927384806198809e-06,
+      "loss": 0.3476,
+      "step": 19630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9264190132987276e-06,
+      "loss": 0.319,
+      "step": 19631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925453284123112e-06,
+      "loss": 0.3241,
+      "step": 19632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.924487618684083e-06,
+      "loss": 0.311,
+      "step": 19633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923522016993777e-06,
+      "loss": 0.299,
+      "step": 19634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.922556479064319e-06,
+      "loss": 0.3057,
+      "step": 19635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.921591004907833e-06,
+      "loss": 0.3243,
+      "step": 19636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920625594536448e-06,
+      "loss": 0.3031,
+      "step": 19637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919660247962294e-06,
+      "loss": 0.3286,
+      "step": 19638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.918694965197487e-06,
+      "loss": 0.3228,
+      "step": 19639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917729746254156e-06,
+      "loss": 0.3306,
+      "step": 19640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9167645911444285e-06,
+      "loss": 0.2974,
+      "step": 19641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915799499880419e-06,
+      "loss": 0.3247,
+      "step": 19642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.914834472474252e-06,
+      "loss": 0.3269,
+      "step": 19643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.913869508938053e-06,
+      "loss": 0.3034,
+      "step": 19644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912904609283937e-06,
+      "loss": 0.3214,
+      "step": 19645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91193977352402e-06,
+      "loss": 0.2801,
+      "step": 19646
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910975001670426e-06,
+      "loss": 0.339,
+      "step": 19647
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910010293735273e-06,
+      "loss": 0.3429,
+      "step": 19648
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.909045649730674e-06,
+      "loss": 0.2882,
+      "step": 19649
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9080810696687455e-06,
+      "loss": 0.3447,
+      "step": 19650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.907116553561608e-06,
+      "loss": 0.3313,
+      "step": 19651
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906152101421368e-06,
+      "loss": 0.2749,
+      "step": 19652
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.905187713260142e-06,
+      "loss": 0.3188,
+      "step": 19653
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9042233890900485e-06,
+      "loss": 0.2972,
+      "step": 19654
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.903259128923195e-06,
+      "loss": 0.3244,
+      "step": 19655
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902294932771687e-06,
+      "loss": 0.3135,
+      "step": 19656
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9013308006476455e-06,
+      "loss": 0.3395,
+      "step": 19657
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.90036673256317e-06,
+      "loss": 0.2984,
+      "step": 19658
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8994027285303745e-06,
+      "loss": 0.325,
+      "step": 19659
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8984387885613695e-06,
+      "loss": 0.3295,
+      "step": 19660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.897474912668255e-06,
+      "loss": 0.3097,
+      "step": 19661
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896511100863142e-06,
+      "loss": 0.3244,
+      "step": 19662
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8955473531581395e-06,
+      "loss": 0.3388,
+      "step": 19663
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.894583669565345e-06,
+      "loss": 0.3188,
+      "step": 19664
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.893620050096868e-06,
+      "loss": 0.2947,
+      "step": 19665
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.892656494764806e-06,
+      "loss": 0.3247,
+      "step": 19666
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891693003581267e-06,
+      "loss": 0.3209,
+      "step": 19667
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890729576558347e-06,
+      "loss": 0.3402,
+      "step": 19668
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889766213708154e-06,
+      "loss": 0.2958,
+      "step": 19669
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88880291504278e-06,
+      "loss": 0.328,
+      "step": 19670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8878396805743264e-06,
+      "loss": 0.2984,
+      "step": 19671
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8868765103148975e-06,
+      "loss": 0.3201,
+      "step": 19672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8859134042765825e-06,
+      "loss": 0.3361,
+      "step": 19673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88495036247148e-06,
+      "loss": 0.3528,
+      "step": 19674
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.883987384911692e-06,
+      "loss": 0.3522,
+      "step": 19675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8830244716093056e-06,
+      "loss": 0.2998,
+      "step": 19676
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882061622576421e-06,
+      "loss": 0.3199,
+      "step": 19677
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881098837825131e-06,
+      "loss": 0.3376,
+      "step": 19678
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880136117367521e-06,
+      "loss": 0.3239,
+      "step": 19679
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879173461215689e-06,
+      "loss": 0.2896,
+      "step": 19680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878210869381729e-06,
+      "loss": 0.317,
+      "step": 19681
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.877248341877723e-06,
+      "loss": 0.3195,
+      "step": 19682
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.876285878715764e-06,
+      "loss": 0.3282,
+      "step": 19683
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875323479907942e-06,
+      "loss": 0.3243,
+      "step": 19684
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.874361145466349e-06,
+      "loss": 0.3217,
+      "step": 19685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8733988754030615e-06,
+      "loss": 0.3101,
+      "step": 19686
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872436669730176e-06,
+      "loss": 0.2789,
+      "step": 19687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8714745284597735e-06,
+      "loss": 0.3025,
+      "step": 19688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.870512451603933e-06,
+      "loss": 0.3071,
+      "step": 19689
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869550439174744e-06,
+      "loss": 0.3331,
+      "step": 19690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8685884911842925e-06,
+      "loss": 0.3068,
+      "step": 19691
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.867626607644654e-06,
+      "loss": 0.3498,
+      "step": 19692
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866664788567912e-06,
+      "loss": 0.2846,
+      "step": 19693
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.865703033966153e-06,
+      "loss": 0.3419,
+      "step": 19694
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.864741343851446e-06,
+      "loss": 0.3194,
+      "step": 19695
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.863779718235875e-06,
+      "loss": 0.2862,
+      "step": 19696
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.862818157131524e-06,
+      "loss": 0.3182,
+      "step": 19697
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.861856660550462e-06,
+      "loss": 0.3525,
+      "step": 19698
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.860895228504766e-06,
+      "loss": 0.322,
+      "step": 19699
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8599338610065185e-06,
+      "loss": 0.3125,
+      "step": 19700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.858972558067784e-06,
+      "loss": 0.3149,
+      "step": 19701
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.858011319700643e-06,
+      "loss": 0.3133,
+      "step": 19702
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857050145917172e-06,
+      "loss": 0.348,
+      "step": 19703
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856089036729435e-06,
+      "loss": 0.2925,
+      "step": 19704
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8551279921495075e-06,
+      "loss": 0.3227,
+      "step": 19705
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.854167012189464e-06,
+      "loss": 0.3258,
+      "step": 19706
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853206096861368e-06,
+      "loss": 0.3487,
+      "step": 19707
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.852245246177295e-06,
+      "loss": 0.322,
+      "step": 19708
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85128446014931e-06,
+      "loss": 0.3315,
+      "step": 19709
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.850323738789477e-06,
+      "loss": 0.299,
+      "step": 19710
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8493630821098655e-06,
+      "loss": 0.3199,
+      "step": 19711
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848402490122548e-06,
+      "loss": 0.3159,
+      "step": 19712
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847441962839579e-06,
+      "loss": 0.321,
+      "step": 19713
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.846481500273026e-06,
+      "loss": 0.313,
+      "step": 19714
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845521102434959e-06,
+      "loss": 0.3391,
+      "step": 19715
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844560769337432e-06,
+      "loss": 0.3155,
+      "step": 19716
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.84360050099251e-06,
+      "loss": 0.3311,
+      "step": 19717
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8426402974122585e-06,
+      "loss": 0.3459,
+      "step": 19718
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.841680158608734e-06,
+      "loss": 0.301,
+      "step": 19719
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.840720084593992e-06,
+      "loss": 0.3226,
+      "step": 19720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839760075380093e-06,
+      "loss": 0.3532,
+      "step": 19721
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838800130979102e-06,
+      "loss": 0.2686,
+      "step": 19722
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.837840251403067e-06,
+      "loss": 0.2993,
+      "step": 19723
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836880436664047e-06,
+      "loss": 0.3109,
+      "step": 19724
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8359206867741005e-06,
+      "loss": 0.3495,
+      "step": 19725
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.834961001745276e-06,
+      "loss": 0.3236,
+      "step": 19726
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.834001381589631e-06,
+      "loss": 0.3346,
+      "step": 19727
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833041826319222e-06,
+      "loss": 0.3212,
+      "step": 19728
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.832082335946097e-06,
+      "loss": 0.3387,
+      "step": 19729
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.831122910482303e-06,
+      "loss": 0.3509,
+      "step": 19730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830163549939899e-06,
+      "loss": 0.358,
+      "step": 19731
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829204254330927e-06,
+      "loss": 0.2979,
+      "step": 19732
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.828245023667438e-06,
+      "loss": 0.3323,
+      "step": 19733
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827285857961486e-06,
+      "loss": 0.3283,
+      "step": 19734
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826326757225109e-06,
+      "loss": 0.3706,
+      "step": 19735
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.825367721470358e-06,
+      "loss": 0.285,
+      "step": 19736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824408750709282e-06,
+      "loss": 0.3315,
+      "step": 19737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.823449844953917e-06,
+      "loss": 0.3327,
+      "step": 19738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.822491004216316e-06,
+      "loss": 0.3501,
+      "step": 19739
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8215322285085196e-06,
+      "loss": 0.3205,
+      "step": 19740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.820573517842564e-06,
+      "loss": 0.3211,
+      "step": 19741
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.819614872230495e-06,
+      "loss": 0.2806,
+      "step": 19742
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818656291684356e-06,
+      "loss": 0.3013,
+      "step": 19743
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8176977762161815e-06,
+      "loss": 0.3138,
+      "step": 19744
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.816739325838012e-06,
+      "loss": 0.3032,
+      "step": 19745
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8157809405618925e-06,
+      "loss": 0.3341,
+      "step": 19746
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.814822620399849e-06,
+      "loss": 0.3258,
+      "step": 19747
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813864365363925e-06,
+      "loss": 0.3006,
+      "step": 19748
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812906175466159e-06,
+      "loss": 0.3462,
+      "step": 19749
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.811948050718581e-06,
+      "loss": 0.3161,
+      "step": 19750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810989991133224e-06,
+      "loss": 0.3524,
+      "step": 19751
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810031996722125e-06,
+      "loss": 0.3118,
+      "step": 19752
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809074067497313e-06,
+      "loss": 0.3041,
+      "step": 19753
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.808116203470819e-06,
+      "loss": 0.3341,
+      "step": 19754
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807158404654682e-06,
+      "loss": 0.3341,
+      "step": 19755
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.806200671060922e-06,
+      "loss": 0.3182,
+      "step": 19756
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805243002701573e-06,
+      "loss": 0.3357,
+      "step": 19757
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.804285399588662e-06,
+      "loss": 0.32,
+      "step": 19758
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80332786173422e-06,
+      "loss": 0.3435,
+      "step": 19759
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.802370389150274e-06,
+      "loss": 0.3151,
+      "step": 19760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801412981848841e-06,
+      "loss": 0.2995,
+      "step": 19761
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800455639841956e-06,
+      "loss": 0.3331,
+      "step": 19762
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799498363141637e-06,
+      "loss": 0.3322,
+      "step": 19763
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.798541151759909e-06,
+      "loss": 0.3268,
+      "step": 19764
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797584005708799e-06,
+      "loss": 0.3026,
+      "step": 19765
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796626925000322e-06,
+      "loss": 0.3121,
+      "step": 19766
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.795669909646501e-06,
+      "loss": 0.3321,
+      "step": 19767
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794712959659362e-06,
+      "loss": 0.3224,
+      "step": 19768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.793756075050916e-06,
+      "loss": 0.2831,
+      "step": 19769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7927992558331895e-06,
+      "loss": 0.3233,
+      "step": 19770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7918425020181915e-06,
+      "loss": 0.3484,
+      "step": 19771
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790885813617948e-06,
+      "loss": 0.3232,
+      "step": 19772
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789929190644466e-06,
+      "loss": 0.3243,
+      "step": 19773
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788972633109768e-06,
+      "loss": 0.3135,
+      "step": 19774
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788016141025864e-06,
+      "loss": 0.3209,
+      "step": 19775
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.787059714404768e-06,
+      "loss": 0.3279,
+      "step": 19776
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.786103353258498e-06,
+      "loss": 0.3236,
+      "step": 19777
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785147057599057e-06,
+      "loss": 0.3333,
+      "step": 19778
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.784190827438462e-06,
+      "loss": 0.2923,
+      "step": 19779
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783234662788726e-06,
+      "loss": 0.3403,
+      "step": 19780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.782278563661851e-06,
+      "loss": 0.3121,
+      "step": 19781
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7813225300698544e-06,
+      "loss": 0.3038,
+      "step": 19782
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780366562024737e-06,
+      "loss": 0.2927,
+      "step": 19783
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779410659538505e-06,
+      "loss": 0.3253,
+      "step": 19784
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.778454822623167e-06,
+      "loss": 0.3259,
+      "step": 19785
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7774990512907335e-06,
+      "loss": 0.3132,
+      "step": 19786
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7765433455532005e-06,
+      "loss": 0.3094,
+      "step": 19787
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7755877054225765e-06,
+      "loss": 0.3486,
+      "step": 19788
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774632130910866e-06,
+      "loss": 0.2854,
+      "step": 19789
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.773676622030064e-06,
+      "loss": 0.3535,
+      "step": 19790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.772721178792178e-06,
+      "loss": 0.3045,
+      "step": 19791
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771765801209211e-06,
+      "loss": 0.3396,
+      "step": 19792
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.770810489293156e-06,
+      "loss": 0.3333,
+      "step": 19793
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769855243056012e-06,
+      "loss": 0.2946,
+      "step": 19794
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.768900062509779e-06,
+      "loss": 0.297,
+      "step": 19795
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.767944947666458e-06,
+      "loss": 0.3103,
+      "step": 19796
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.766989898538037e-06,
+      "loss": 0.3577,
+      "step": 19797
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7660349151365175e-06,
+      "loss": 0.3021,
+      "step": 19798
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765079997473895e-06,
+      "loss": 0.3228,
+      "step": 19799
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764125145562156e-06,
+      "loss": 0.2878,
+      "step": 19800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7631703594133005e-06,
+      "loss": 0.3109,
+      "step": 19801
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762215639039321e-06,
+      "loss": 0.3279,
+      "step": 19802
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7612609844522076e-06,
+      "loss": 0.3576,
+      "step": 19803
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760306395663945e-06,
+      "loss": 0.3024,
+      "step": 19804
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759351872686532e-06,
+      "loss": 0.2893,
+      "step": 19805
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758397415531949e-06,
+      "loss": 0.3442,
+      "step": 19806
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.757443024212188e-06,
+      "loss": 0.3271,
+      "step": 19807
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75648869873924e-06,
+      "loss": 0.3526,
+      "step": 19808
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.755534439125084e-06,
+      "loss": 0.3096,
+      "step": 19809
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754580245381708e-06,
+      "loss": 0.3228,
+      "step": 19810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753626117521103e-06,
+      "loss": 0.3443,
+      "step": 19811
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.752672055555244e-06,
+      "loss": 0.3475,
+      "step": 19812
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751718059496121e-06,
+      "loss": 0.2928,
+      "step": 19813
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7507641293557124e-06,
+      "loss": 0.3012,
+      "step": 19814
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.749810265145999e-06,
+      "loss": 0.2834,
+      "step": 19815
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.748856466878961e-06,
+      "loss": 0.3225,
+      "step": 19816
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747902734566583e-06,
+      "loss": 0.3619,
+      "step": 19817
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.746949068220838e-06,
+      "loss": 0.2996,
+      "step": 19818
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7459954678537064e-06,
+      "loss": 0.3223,
+      "step": 19819
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74504193347717e-06,
+      "loss": 0.3253,
+      "step": 19820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.744088465103198e-06,
+      "loss": 0.3291,
+      "step": 19821
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7431350627437685e-06,
+      "loss": 0.3347,
+      "step": 19822
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742181726410862e-06,
+      "loss": 0.3273,
+      "step": 19823
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741228456116447e-06,
+      "loss": 0.3374,
+      "step": 19824
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7402752518724925e-06,
+      "loss": 0.297,
+      "step": 19825
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7393221136909796e-06,
+      "loss": 0.3368,
+      "step": 19826
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738369041583871e-06,
+      "loss": 0.3335,
+      "step": 19827
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7374160355631435e-06,
+      "loss": 0.3082,
+      "step": 19828
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736463095640764e-06,
+      "loss": 0.3051,
+      "step": 19829
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.735510221828708e-06,
+      "loss": 0.3524,
+      "step": 19830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.734557414138934e-06,
+      "loss": 0.3696,
+      "step": 19831
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7336046725834125e-06,
+      "loss": 0.3544,
+      "step": 19832
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732651997174117e-06,
+      "loss": 0.3298,
+      "step": 19833
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731699387923006e-06,
+      "loss": 0.3682,
+      "step": 19834
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7307468448420435e-06,
+      "loss": 0.2827,
+      "step": 19835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.729794367943198e-06,
+      "loss": 0.3183,
+      "step": 19836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728841957238427e-06,
+      "loss": 0.3462,
+      "step": 19837
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.727889612739697e-06,
+      "loss": 0.308,
+      "step": 19838
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726937334458972e-06,
+      "loss": 0.2918,
+      "step": 19839
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.725985122408206e-06,
+      "loss": 0.3243,
+      "step": 19840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.725032976599362e-06,
+      "loss": 0.3183,
+      "step": 19841
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.724080897044402e-06,
+      "loss": 0.3392,
+      "step": 19842
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723128883755279e-06,
+      "loss": 0.3325,
+      "step": 19843
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722176936743956e-06,
+      "loss": 0.3498,
+      "step": 19844
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.721225056022383e-06,
+      "loss": 0.3229,
+      "step": 19845
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720273241602522e-06,
+      "loss": 0.3405,
+      "step": 19846
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7193214934963204e-06,
+      "loss": 0.3255,
+      "step": 19847
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7183698117157414e-06,
+      "loss": 0.3365,
+      "step": 19848
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71741819627273e-06,
+      "loss": 0.3513,
+      "step": 19849
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.716466647179241e-06,
+      "loss": 0.2991,
+      "step": 19850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71551516444723e-06,
+      "loss": 0.3463,
+      "step": 19851
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.714563748088643e-06,
+      "loss": 0.3223,
+      "step": 19852
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713612398115429e-06,
+      "loss": 0.321,
+      "step": 19853
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.712661114539545e-06,
+      "loss": 0.2923,
+      "step": 19854
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711709897372928e-06,
+      "loss": 0.3071,
+      "step": 19855
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.710758746627536e-06,
+      "loss": 0.3263,
+      "step": 19856
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70980766231531e-06,
+      "loss": 0.3415,
+      "step": 19857
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708856644448192e-06,
+      "loss": 0.2913,
+      "step": 19858
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7079056930381305e-06,
+      "loss": 0.3294,
+      "step": 19859
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7069548080970754e-06,
+      "loss": 0.3163,
+      "step": 19860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.706003989636959e-06,
+      "loss": 0.3426,
+      "step": 19861
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.705053237669729e-06,
+      "loss": 0.3311,
+      "step": 19862
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.704102552207331e-06,
+      "loss": 0.3234,
+      "step": 19863
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703151933261698e-06,
+      "loss": 0.3199,
+      "step": 19864
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.702201380844773e-06,
+      "loss": 0.319,
+      "step": 19865
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7012508949685e-06,
+      "loss": 0.3191,
+      "step": 19866
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700300475644811e-06,
+      "loss": 0.3473,
+      "step": 19867
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699350122885642e-06,
+      "loss": 0.3406,
+      "step": 19868
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698399836702933e-06,
+      "loss": 0.34,
+      "step": 19869
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697449617108622e-06,
+      "loss": 0.283,
+      "step": 19870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696499464114637e-06,
+      "loss": 0.3353,
+      "step": 19871
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695549377732916e-06,
+      "loss": 0.3177,
+      "step": 19872
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.694599357975395e-06,
+      "loss": 0.3049,
+      "step": 19873
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.693649404853999e-06,
+      "loss": 0.2944,
+      "step": 19874
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.692699518380664e-06,
+      "loss": 0.3251,
+      "step": 19875
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.691749698567325e-06,
+      "loss": 0.2979,
+      "step": 19876
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690799945425907e-06,
+      "loss": 0.3115,
+      "step": 19877
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.689850258968334e-06,
+      "loss": 0.328,
+      "step": 19878
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688900639206545e-06,
+      "loss": 0.3053,
+      "step": 19879
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687951086152455e-06,
+      "loss": 0.3359,
+      "step": 19880
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687001599818e-06,
+      "loss": 0.3118,
+      "step": 19881
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.686052180215104e-06,
+      "loss": 0.3308,
+      "step": 19882
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685102827355687e-06,
+      "loss": 0.2932,
+      "step": 19883
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.684153541251676e-06,
+      "loss": 0.326,
+      "step": 19884
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.683204321914998e-06,
+      "loss": 0.3156,
+      "step": 19885
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.682255169357568e-06,
+      "loss": 0.2923,
+      "step": 19886
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681306083591315e-06,
+      "loss": 0.3331,
+      "step": 19887
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.680357064628155e-06,
+      "loss": 0.3438,
+      "step": 19888
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679408112480005e-06,
+      "loss": 0.3176,
+      "step": 19889
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.678459227158786e-06,
+      "loss": 0.3427,
+      "step": 19890
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.677510408676422e-06,
+      "loss": 0.3233,
+      "step": 19891
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.676561657044821e-06,
+      "loss": 0.3135,
+      "step": 19892
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675612972275904e-06,
+      "loss": 0.2999,
+      "step": 19893
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.67466435438159e-06,
+      "loss": 0.3262,
+      "step": 19894
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673715803373786e-06,
+      "loss": 0.3242,
+      "step": 19895
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.67276731926441e-06,
+      "loss": 0.2921,
+      "step": 19896
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.671818902065378e-06,
+      "loss": 0.3303,
+      "step": 19897
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6708705517885996e-06,
+      "loss": 0.2946,
+      "step": 19898
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669922268445981e-06,
+      "loss": 0.288,
+      "step": 19899
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.668974052049442e-06,
+      "loss": 0.3026,
+      "step": 19900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.668025902610883e-06,
+      "loss": 0.2985,
+      "step": 19901
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.667077820142219e-06,
+      "loss": 0.3289,
+      "step": 19902
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.666129804655354e-06,
+      "loss": 0.3192,
+      "step": 19903
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665181856162202e-06,
+      "loss": 0.3254,
+      "step": 19904
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.664233974674661e-06,
+      "loss": 0.3067,
+      "step": 19905
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6632861602046394e-06,
+      "loss": 0.3234,
+      "step": 19906
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.662338412764048e-06,
+      "loss": 0.3514,
+      "step": 19907
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.661390732364783e-06,
+      "loss": 0.3719,
+      "step": 19908
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.660443119018747e-06,
+      "loss": 0.3458,
+      "step": 19909
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659495572737849e-06,
+      "loss": 0.3033,
+      "step": 19910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65854809353398e-06,
+      "loss": 0.3085,
+      "step": 19911
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657600681419047e-06,
+      "loss": 0.3469,
+      "step": 19912
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.656653336404953e-06,
+      "loss": 0.3246,
+      "step": 19913
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655706058503587e-06,
+      "loss": 0.2901,
+      "step": 19914
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654758847726854e-06,
+      "loss": 0.3642,
+      "step": 19915
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653811704086652e-06,
+      "loss": 0.3117,
+      "step": 19916
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652864627594871e-06,
+      "loss": 0.2998,
+      "step": 19917
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651917618263413e-06,
+      "loss": 0.3244,
+      "step": 19918
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65097067610417e-06,
+      "loss": 0.2976,
+      "step": 19919
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65002380112903e-06,
+      "loss": 0.3452,
+      "step": 19920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6490769933498915e-06,
+      "loss": 0.3139,
+      "step": 19921
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64813025277865e-06,
+      "loss": 0.3109,
+      "step": 19922
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647183579427187e-06,
+      "loss": 0.3065,
+      "step": 19923
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.646236973307398e-06,
+      "loss": 0.3098,
+      "step": 19924
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.645290434431178e-06,
+      "loss": 0.3454,
+      "step": 19925
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.644343962810406e-06,
+      "loss": 0.3327,
+      "step": 19926
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643397558456973e-06,
+      "loss": 0.2826,
+      "step": 19927
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64245122138277e-06,
+      "loss": 0.3243,
+      "step": 19928
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64150495159968e-06,
+      "loss": 0.3028,
+      "step": 19929
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.640558749119584e-06,
+      "loss": 0.3046,
+      "step": 19930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639612613954375e-06,
+      "loss": 0.3105,
+      "step": 19931
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.638666546115927e-06,
+      "loss": 0.3136,
+      "step": 19932
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637720545616129e-06,
+      "loss": 0.3124,
+      "step": 19933
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636774612466863e-06,
+      "loss": 0.3198,
+      "step": 19934
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635828746680005e-06,
+      "loss": 0.2976,
+      "step": 19935
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6348829482674395e-06,
+      "loss": 0.3405,
+      "step": 19936
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.633937217241048e-06,
+      "loss": 0.3489,
+      "step": 19937
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632991553612701e-06,
+      "loss": 0.3747,
+      "step": 19938
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632045957394286e-06,
+      "loss": 0.2943,
+      "step": 19939
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.631100428597669e-06,
+      "loss": 0.316,
+      "step": 19940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630154967234736e-06,
+      "loss": 0.3193,
+      "step": 19941
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629209573317352e-06,
+      "loss": 0.3187,
+      "step": 19942
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628264246857399e-06,
+      "loss": 0.3419,
+      "step": 19943
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62731898786675e-06,
+      "loss": 0.2824,
+      "step": 19944
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626373796357273e-06,
+      "loss": 0.2975,
+      "step": 19945
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625428672340841e-06,
+      "loss": 0.278,
+      "step": 19946
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624483615829329e-06,
+      "loss": 0.3177,
+      "step": 19947
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623538626834602e-06,
+      "loss": 0.3473,
+      "step": 19948
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.622593705368533e-06,
+      "loss": 0.291,
+      "step": 19949
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621648851442986e-06,
+      "loss": 0.3217,
+      "step": 19950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620704065069833e-06,
+      "loss": 0.2991,
+      "step": 19951
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619759346260934e-06,
+      "loss": 0.3262,
+      "step": 19952
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618814695028164e-06,
+      "loss": 0.3277,
+      "step": 19953
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.617870111383379e-06,
+      "loss": 0.3308,
+      "step": 19954
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.616925595338448e-06,
+      "loss": 0.3154,
+      "step": 19955
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615981146905234e-06,
+      "loss": 0.3073,
+      "step": 19956
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615036766095597e-06,
+      "loss": 0.3147,
+      "step": 19957
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614092452921398e-06,
+      "loss": 0.3152,
+      "step": 19958
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613148207394505e-06,
+      "loss": 0.2754,
+      "step": 19959
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612204029526767e-06,
+      "loss": 0.3498,
+      "step": 19960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6112599193300525e-06,
+      "loss": 0.3221,
+      "step": 19961
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.610315876816215e-06,
+      "loss": 0.3553,
+      "step": 19962
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609371901997108e-06,
+      "loss": 0.3394,
+      "step": 19963
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608427994884592e-06,
+      "loss": 0.3354,
+      "step": 19964
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607484155490526e-06,
+      "loss": 0.2975,
+      "step": 19965
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606540383826758e-06,
+      "loss": 0.3142,
+      "step": 19966
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6055966799051436e-06,
+      "loss": 0.2918,
+      "step": 19967
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.604653043737541e-06,
+      "loss": 0.2935,
+      "step": 19968
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6037094753357954e-06,
+      "loss": 0.3374,
+      "step": 19969
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602765974711761e-06,
+      "loss": 0.3031,
+      "step": 19970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.601822541877291e-06,
+      "loss": 0.3224,
+      "step": 19971
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600879176844233e-06,
+      "loss": 0.3079,
+      "step": 19972
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.599935879624431e-06,
+      "loss": 0.3155,
+      "step": 19973
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598992650229741e-06,
+      "loss": 0.3219,
+      "step": 19974
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598049488672001e-06,
+      "loss": 0.2977,
+      "step": 19975
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597106394963063e-06,
+      "loss": 0.3024,
+      "step": 19976
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59616336911477e-06,
+      "loss": 0.2859,
+      "step": 19977
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.595220411138973e-06,
+      "loss": 0.2889,
+      "step": 19978
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594277521047505e-06,
+      "loss": 0.3226,
+      "step": 19979
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593334698852215e-06,
+      "loss": 0.3444,
+      "step": 19980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.592391944564949e-06,
+      "loss": 0.3268,
+      "step": 19981
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591449258197541e-06,
+      "loss": 0.2841,
+      "step": 19982
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59050663976183e-06,
+      "loss": 0.3035,
+      "step": 19983
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.589564089269663e-06,
+      "loss": 0.2965,
+      "step": 19984
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.58862160673287e-06,
+      "loss": 0.3355,
+      "step": 19985
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.587679192163293e-06,
+      "loss": 0.2712,
+      "step": 19986
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5867368455727715e-06,
+      "loss": 0.3161,
+      "step": 19987
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.585794566973134e-06,
+      "loss": 0.3226,
+      "step": 19988
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.584852356376222e-06,
+      "loss": 0.3271,
+      "step": 19989
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.58391021379387e-06,
+      "loss": 0.3444,
+      "step": 19990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582968139237906e-06,
+      "loss": 0.3425,
+      "step": 19991
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5820261327201684e-06,
+      "loss": 0.322,
+      "step": 19992
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.581084194252486e-06,
+      "loss": 0.3037,
+      "step": 19993
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580142323846687e-06,
+      "loss": 0.3277,
+      "step": 19994
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.579200521514603e-06,
+      "loss": 0.3367,
+      "step": 19995
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5782587872680685e-06,
+      "loss": 0.3176,
+      "step": 19996
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577317121118903e-06,
+      "loss": 0.3148,
+      "step": 19997
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5763755230789386e-06,
+      "loss": 0.2652,
+      "step": 19998
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.575433993160005e-06,
+      "loss": 0.3364,
+      "step": 19999
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574492531373921e-06,
+      "loss": 0.3154,
+      "step": 20000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.573551137732514e-06,
+      "loss": 0.309,
+      "step": 20001
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572609812247611e-06,
+      "loss": 0.3304,
+      "step": 20002
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571668554931035e-06,
+      "loss": 0.2839,
+      "step": 20003
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.570727365794602e-06,
+      "loss": 0.2933,
+      "step": 20004
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569786244850139e-06,
+      "loss": 0.3022,
+      "step": 20005
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5688451921094636e-06,
+      "loss": 0.3382,
+      "step": 20006
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.567904207584396e-06,
+      "loss": 0.3275,
+      "step": 20007
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.566963291286759e-06,
+      "loss": 0.313,
+      "step": 20008
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.566022443228364e-06,
+      "loss": 0.3293,
+      "step": 20009
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565081663421032e-06,
+      "loss": 0.2948,
+      "step": 20010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.564140951876581e-06,
+      "loss": 0.3548,
+      "step": 20011
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.563200308606821e-06,
+      "loss": 0.35,
+      "step": 20012
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5622597336235745e-06,
+      "loss": 0.3217,
+      "step": 20013
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.561319226938645e-06,
+      "loss": 0.3292,
+      "step": 20014
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.560378788563855e-06,
+      "loss": 0.3314,
+      "step": 20015
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559438418511008e-06,
+      "loss": 0.3154,
+      "step": 20016
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.55849811679192e-06,
+      "loss": 0.3561,
+      "step": 20017
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557557883418403e-06,
+      "loss": 0.3033,
+      "step": 20018
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556617718402261e-06,
+      "loss": 0.3134,
+      "step": 20019
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5556776217553055e-06,
+      "loss": 0.3429,
+      "step": 20020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554737593489347e-06,
+      "loss": 0.3483,
+      "step": 20021
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553797633616186e-06,
+      "loss": 0.3002,
+      "step": 20022
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552857742147635e-06,
+      "loss": 0.3235,
+      "step": 20023
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551917919095495e-06,
+      "loss": 0.3503,
+      "step": 20024
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550978164471569e-06,
+      "loss": 0.3104,
+      "step": 20025
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550038478287661e-06,
+      "loss": 0.2894,
+      "step": 20026
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549098860555581e-06,
+      "loss": 0.3319,
+      "step": 20027
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.548159311287119e-06,
+      "loss": 0.288,
+      "step": 20028
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5472198304940815e-06,
+      "loss": 0.3188,
+      "step": 20029
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5462804181882725e-06,
+      "loss": 0.3169,
+      "step": 20030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545341074381483e-06,
+      "loss": 0.3138,
+      "step": 20031
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544401799085515e-06,
+      "loss": 0.3202,
+      "step": 20032
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54346259231217e-06,
+      "loss": 0.2736,
+      "step": 20033
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54252345407324e-06,
+      "loss": 0.3077,
+      "step": 20034
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541584384380518e-06,
+      "loss": 0.2992,
+      "step": 20035
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.540645383245804e-06,
+      "loss": 0.3336,
+      "step": 20036
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.539706450680888e-06,
+      "loss": 0.2877,
+      "step": 20037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.538767586697563e-06,
+      "loss": 0.3251,
+      "step": 20038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.537828791307628e-06,
+      "loss": 0.351,
+      "step": 20039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536890064522865e-06,
+      "loss": 0.3262,
+      "step": 20040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.535951406355069e-06,
+      "loss": 0.2899,
+      "step": 20041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.535012816816034e-06,
+      "loss": 0.3033,
+      "step": 20042
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.534074295917539e-06,
+      "loss": 0.3278,
+      "step": 20043
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53313584367138e-06,
+      "loss": 0.3534,
+      "step": 20044
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.532197460089339e-06,
+      "loss": 0.3005,
+      "step": 20045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531259145183207e-06,
+      "loss": 0.3389,
+      "step": 20046
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530320898964763e-06,
+      "loss": 0.3175,
+      "step": 20047
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.529382721445796e-06,
+      "loss": 0.2823,
+      "step": 20048
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52844461263809e-06,
+      "loss": 0.2669,
+      "step": 20049
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5275065725534225e-06,
+      "loss": 0.3449,
+      "step": 20050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52656860120358e-06,
+      "loss": 0.3031,
+      "step": 20051
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525630698600346e-06,
+      "loss": 0.3286,
+      "step": 20052
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524692864755494e-06,
+      "loss": 0.3314,
+      "step": 20053
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5237550996808075e-06,
+      "loss": 0.3426,
+      "step": 20054
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522817403388062e-06,
+      "loss": 0.3031,
+      "step": 20055
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52187977588904e-06,
+      "loss": 0.3529,
+      "step": 20056
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520942217195511e-06,
+      "loss": 0.286,
+      "step": 20057
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520004727319258e-06,
+      "loss": 0.3325,
+      "step": 20058
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.51906730627205e-06,
+      "loss": 0.3216,
+      "step": 20059
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518129954065662e-06,
+      "loss": 0.3387,
+      "step": 20060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.517192670711874e-06,
+      "loss": 0.2702,
+      "step": 20061
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516255456222448e-06,
+      "loss": 0.2731,
+      "step": 20062
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.515318310609161e-06,
+      "loss": 0.3364,
+      "step": 20063
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.514381233883788e-06,
+      "loss": 0.3032,
+      "step": 20064
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513444226058089e-06,
+      "loss": 0.3103,
+      "step": 20065
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5125072871438426e-06,
+      "loss": 0.3369,
+      "step": 20066
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5115704171528105e-06,
+      "loss": 0.3084,
+      "step": 20067
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510633616096759e-06,
+      "loss": 0.3217,
+      "step": 20068
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509696883987457e-06,
+      "loss": 0.3412,
+      "step": 20069
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508760220836672e-06,
+      "loss": 0.3342,
+      "step": 20070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507823626656163e-06,
+      "loss": 0.3206,
+      "step": 20071
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5068871014576965e-06,
+      "loss": 0.3156,
+      "step": 20072
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.50595064525304e-06,
+      "loss": 0.3169,
+      "step": 20073
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505014258053947e-06,
+      "loss": 0.3234,
+      "step": 20074
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504077939872183e-06,
+      "loss": 0.3103,
+      "step": 20075
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.503141690719512e-06,
+      "loss": 0.3516,
+      "step": 20076
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.502205510607688e-06,
+      "loss": 0.2892,
+      "step": 20077
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501269399548468e-06,
+      "loss": 0.335,
+      "step": 20078
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.500333357553615e-06,
+      "loss": 0.3495,
+      "step": 20079
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.49939738463488e-06,
+      "loss": 0.3365,
+      "step": 20080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498461480804022e-06,
+      "loss": 0.3359,
+      "step": 20081
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.497525646072799e-06,
+      "loss": 0.3132,
+      "step": 20082
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496589880452959e-06,
+      "loss": 0.321,
+      "step": 20083
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495654183956258e-06,
+      "loss": 0.3514,
+      "step": 20084
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49471855659445e-06,
+      "loss": 0.3053,
+      "step": 20085
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4937829983792865e-06,
+      "loss": 0.3534,
+      "step": 20086
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.492847509322519e-06,
+      "loss": 0.301,
+      "step": 20087
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49191208943589e-06,
+      "loss": 0.3036,
+      "step": 20088
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.490976738731159e-06,
+      "loss": 0.3708,
+      "step": 20089
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.490041457220064e-06,
+      "loss": 0.3124,
+      "step": 20090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.489106244914358e-06,
+      "loss": 0.3163,
+      "step": 20091
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.488171101825788e-06,
+      "loss": 0.3095,
+      "step": 20092
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.487236027966096e-06,
+      "loss": 0.3467,
+      "step": 20093
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.486301023347026e-06,
+      "loss": 0.3527,
+      "step": 20094
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.485366087980329e-06,
+      "loss": 0.2928,
+      "step": 20095
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484431221877739e-06,
+      "loss": 0.3106,
+      "step": 20096
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4834964250510045e-06,
+      "loss": 0.3442,
+      "step": 20097
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.482561697511865e-06,
+      "loss": 0.3261,
+      "step": 20098
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481627039272056e-06,
+      "loss": 0.3375,
+      "step": 20099
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48069245034332e-06,
+      "loss": 0.3425,
+      "step": 20100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479757930737399e-06,
+      "loss": 0.3384,
+      "step": 20101
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478823480466025e-06,
+      "loss": 0.3026,
+      "step": 20102
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477889099540937e-06,
+      "loss": 0.3088,
+      "step": 20103
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.476954787973875e-06,
+      "loss": 0.3204,
+      "step": 20104
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.476020545776566e-06,
+      "loss": 0.2989,
+      "step": 20105
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475086372960749e-06,
+      "loss": 0.3377,
+      "step": 20106
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474152269538159e-06,
+      "loss": 0.3407,
+      "step": 20107
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473218235520526e-06,
+      "loss": 0.3515,
+      "step": 20108
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.472284270919579e-06,
+      "loss": 0.3364,
+      "step": 20109
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4713503757470544e-06,
+      "loss": 0.3199,
+      "step": 20110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.470416550014675e-06,
+      "loss": 0.318,
+      "step": 20111
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469482793734174e-06,
+      "loss": 0.3415,
+      "step": 20112
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.468549106917281e-06,
+      "loss": 0.3244,
+      "step": 20113
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46761548957572e-06,
+      "loss": 0.3089,
+      "step": 20114
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466681941721216e-06,
+      "loss": 0.3187,
+      "step": 20115
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465748463365501e-06,
+      "loss": 0.3209,
+      "step": 20116
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4648150545202915e-06,
+      "loss": 0.337,
+      "step": 20117
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463881715197319e-06,
+      "loss": 0.3294,
+      "step": 20118
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.462948445408298e-06,
+      "loss": 0.3246,
+      "step": 20119
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.462015245164959e-06,
+      "loss": 0.3365,
+      "step": 20120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461082114479014e-06,
+      "loss": 0.32,
+      "step": 20121
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4601490533621886e-06,
+      "loss": 0.3184,
+      "step": 20122
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4592160618262046e-06,
+      "loss": 0.3312,
+      "step": 20123
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4582831398827744e-06,
+      "loss": 0.3296,
+      "step": 20124
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457350287543617e-06,
+      "loss": 0.3295,
+      "step": 20125
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.456417504820455e-06,
+      "loss": 0.3308,
+      "step": 20126
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455484791724995e-06,
+      "loss": 0.3223,
+      "step": 20127
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454552148268961e-06,
+      "loss": 0.3015,
+      "step": 20128
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45361957446406e-06,
+      "loss": 0.3172,
+      "step": 20129
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45268707032201e-06,
+      "loss": 0.2786,
+      "step": 20130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4517546358545175e-06,
+      "loss": 0.3148,
+      "step": 20131
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.450822271073302e-06,
+      "loss": 0.3029,
+      "step": 20132
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.449889975990065e-06,
+      "loss": 0.324,
+      "step": 20133
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44895775061652e-06,
+      "loss": 0.3546,
+      "step": 20134
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4480255949643825e-06,
+      "loss": 0.2946,
+      "step": 20135
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4470935090453495e-06,
+      "loss": 0.32,
+      "step": 20136
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446161492871133e-06,
+      "loss": 0.3041,
+      "step": 20137
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.445229546453443e-06,
+      "loss": 0.3305,
+      "step": 20138
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.2996,
+      "step": 20139
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443365862934448e-06,
+      "loss": 0.3187,
+      "step": 20140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442434125856554e-06,
+      "loss": 0.3197,
+      "step": 20141
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4415024585819955e-06,
+      "loss": 0.3331,
+      "step": 20142
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440570861122476e-06,
+      "loss": 0.3035,
+      "step": 20143
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439639333489704e-06,
+      "loss": 0.3279,
+      "step": 20144
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4387078756953686e-06,
+      "loss": 0.3216,
+      "step": 20145
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437776487751173e-06,
+      "loss": 0.3107,
+      "step": 20146
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436845169668821e-06,
+      "loss": 0.3201,
+      "step": 20147
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.43591392146e-06,
+      "loss": 0.3298,
+      "step": 20148
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434982743136417e-06,
+      "loss": 0.3611,
+      "step": 20149
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434051634709759e-06,
+      "loss": 0.2958,
+      "step": 20150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4331205961917265e-06,
+      "loss": 0.3203,
+      "step": 20151
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4321896275940075e-06,
+      "loss": 0.3262,
+      "step": 20152
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.431258728928304e-06,
+      "loss": 0.3202,
+      "step": 20153
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430327900206297e-06,
+      "loss": 0.3369,
+      "step": 20154
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4293971414396855e-06,
+      "loss": 0.3511,
+      "step": 20155
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.42846645264016e-06,
+      "loss": 0.2878,
+      "step": 20156
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.427535833819404e-06,
+      "loss": 0.3365,
+      "step": 20157
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426605284989112e-06,
+      "loss": 0.3107,
+      "step": 20158
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.425674806160971e-06,
+      "loss": 0.3335,
+      "step": 20159
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4247443973466645e-06,
+      "loss": 0.2881,
+      "step": 20160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423814058557884e-06,
+      "loss": 0.3001,
+      "step": 20161
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.422883789806307e-06,
+      "loss": 0.3174,
+      "step": 20162
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421953591103627e-06,
+      "loss": 0.3081,
+      "step": 20163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421023462461518e-06,
+      "loss": 0.3029,
+      "step": 20164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420093403891666e-06,
+      "loss": 0.3288,
+      "step": 20165
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4191634154057575e-06,
+      "loss": 0.336,
+      "step": 20166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.418233497015465e-06,
+      "loss": 0.3025,
+      "step": 20167
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417303648732473e-06,
+      "loss": 0.3217,
+      "step": 20168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.416373870568463e-06,
+      "loss": 0.3174,
+      "step": 20169
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.415444162535107e-06,
+      "loss": 0.3027,
+      "step": 20170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414514524644088e-06,
+      "loss": 0.3358,
+      "step": 20171
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4135849569070796e-06,
+      "loss": 0.3323,
+      "step": 20172
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412655459335754e-06,
+      "loss": 0.3014,
+      "step": 20173
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411726031941788e-06,
+      "loss": 0.339,
+      "step": 20174
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.41079667473686e-06,
+      "loss": 0.3177,
+      "step": 20175
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409867387732635e-06,
+      "loss": 0.3073,
+      "step": 20176
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408938170940789e-06,
+      "loss": 0.3223,
+      "step": 20177
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408009024372996e-06,
+      "loss": 0.2885,
+      "step": 20178
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.407079948040919e-06,
+      "loss": 0.3531,
+      "step": 20179
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.406150941956231e-06,
+      "loss": 0.2669,
+      "step": 20180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.405222006130605e-06,
+      "loss": 0.3239,
+      "step": 20181
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.404293140575704e-06,
+      "loss": 0.3315,
+      "step": 20182
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403364345303189e-06,
+      "loss": 0.3277,
+      "step": 20183
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402435620324737e-06,
+      "loss": 0.3154,
+      "step": 20184
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.401506965652003e-06,
+      "loss": 0.3282,
+      "step": 20185
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400578381296654e-06,
+      "loss": 0.3204,
+      "step": 20186
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399649867270359e-06,
+      "loss": 0.3485,
+      "step": 20187
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.398721423584771e-06,
+      "loss": 0.306,
+      "step": 20188
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3977930502515566e-06,
+      "loss": 0.2881,
+      "step": 20189
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.396864747282377e-06,
+      "loss": 0.3343,
+      "step": 20190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3959365146888875e-06,
+      "loss": 0.3149,
+      "step": 20191
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.395008352482754e-06,
+      "loss": 0.3076,
+      "step": 20192
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394080260675627e-06,
+      "loss": 0.3378,
+      "step": 20193
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.393152239279165e-06,
+      "loss": 0.3065,
+      "step": 20194
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3922242883050226e-06,
+      "loss": 0.2886,
+      "step": 20195
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391296407764858e-06,
+      "loss": 0.3044,
+      "step": 20196
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.390368597670329e-06,
+      "loss": 0.3443,
+      "step": 20197
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3894408580330805e-06,
+      "loss": 0.2976,
+      "step": 20198
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388513188864768e-06,
+      "loss": 0.308,
+      "step": 20199
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.387585590177049e-06,
+      "loss": 0.339,
+      "step": 20200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386658061981566e-06,
+      "loss": 0.363,
+      "step": 20201
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.385730604289975e-06,
+      "loss": 0.3504,
+      "step": 20202
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.384803217113922e-06,
+      "loss": 0.3414,
+      "step": 20203
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3838759004650515e-06,
+      "loss": 0.2982,
+      "step": 20204
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382948654355014e-06,
+      "loss": 0.3352,
+      "step": 20205
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3820214787954596e-06,
+      "loss": 0.319,
+      "step": 20206
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.381094373798027e-06,
+      "loss": 0.3132,
+      "step": 20207
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380167339374363e-06,
+      "loss": 0.3544,
+      "step": 20208
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379240375536117e-06,
+      "loss": 0.301,
+      "step": 20209
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378313482294922e-06,
+      "loss": 0.3112,
+      "step": 20210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3773866596624244e-06,
+      "loss": 0.3136,
+      "step": 20211
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3764599076502705e-06,
+      "loss": 0.3286,
+      "step": 20212
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375533226270093e-06,
+      "loss": 0.3355,
+      "step": 20213
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3746066155335295e-06,
+      "loss": 0.3342,
+      "step": 20214
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373680075452227e-06,
+      "loss": 0.3192,
+      "step": 20215
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.372753606037814e-06,
+      "loss": 0.3411,
+      "step": 20216
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.371827207301931e-06,
+      "loss": 0.3101,
+      "step": 20217
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.370900879256216e-06,
+      "loss": 0.3147,
+      "step": 20218
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369974621912298e-06,
+      "loss": 0.3193,
+      "step": 20219
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369048435281814e-06,
+      "loss": 0.354,
+      "step": 20220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368122319376402e-06,
+      "loss": 0.2855,
+      "step": 20221
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367196274207684e-06,
+      "loss": 0.3087,
+      "step": 20222
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.366270299787301e-06,
+      "loss": 0.2866,
+      "step": 20223
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365344396126874e-06,
+      "loss": 0.2996,
+      "step": 20224
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364418563238041e-06,
+      "loss": 0.3321,
+      "step": 20225
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363492801132422e-06,
+      "loss": 0.3524,
+      "step": 20226
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362567109821654e-06,
+      "loss": 0.3101,
+      "step": 20227
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.361641489317354e-06,
+      "loss": 0.3442,
+      "step": 20228
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360715939631154e-06,
+      "loss": 0.3041,
+      "step": 20229
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35979046077468e-06,
+      "loss": 0.3132,
+      "step": 20230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358865052759551e-06,
+      "loss": 0.2661,
+      "step": 20231
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.357939715597392e-06,
+      "loss": 0.3094,
+      "step": 20232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35701444929983e-06,
+      "loss": 0.2869,
+      "step": 20233
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3560892538784794e-06,
+      "loss": 0.3189,
+      "step": 20234
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.355164129344968e-06,
+      "loss": 0.2995,
+      "step": 20235
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3542390757109064e-06,
+      "loss": 0.3298,
+      "step": 20236
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353314092987923e-06,
+      "loss": 0.3365,
+      "step": 20237
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.352389181187626e-06,
+      "loss": 0.2936,
+      "step": 20238
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.351464340321638e-06,
+      "loss": 0.3219,
+      "step": 20239
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350539570401578e-06,
+      "loss": 0.3475,
+      "step": 20240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.349614871439053e-06,
+      "loss": 0.3135,
+      "step": 20241
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348690243445682e-06,
+      "loss": 0.3141,
+      "step": 20242
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3477656864330815e-06,
+      "loss": 0.3373,
+      "step": 20243
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.346841200412858e-06,
+      "loss": 0.3087,
+      "step": 20244
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345916785396629e-06,
+      "loss": 0.3196,
+      "step": 20245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344992441396001e-06,
+      "loss": 0.3377,
+      "step": 20246
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344068168422581e-06,
+      "loss": 0.3393,
+      "step": 20247
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343143966487981e-06,
+      "loss": 0.3129,
+      "step": 20248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342219835603815e-06,
+      "loss": 0.3326,
+      "step": 20249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.34129577578168e-06,
+      "loss": 0.3142,
+      "step": 20250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340371787033187e-06,
+      "loss": 0.2542,
+      "step": 20251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339447869369946e-06,
+      "loss": 0.3265,
+      "step": 20252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3385240228035515e-06,
+      "loss": 0.3541,
+      "step": 20253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3376002473456135e-06,
+      "loss": 0.3471,
+      "step": 20254
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.336676543007737e-06,
+      "loss": 0.3599,
+      "step": 20255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.33575290980152e-06,
+      "loss": 0.2968,
+      "step": 20256
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3348293477385605e-06,
+      "loss": 0.3312,
+      "step": 20257
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.333905856830465e-06,
+      "loss": 0.3215,
+      "step": 20258
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332982437088825e-06,
+      "loss": 0.2835,
+      "step": 20259
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3320590885252445e-06,
+      "loss": 0.3236,
+      "step": 20260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331135811151321e-06,
+      "loss": 0.3194,
+      "step": 20261
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3302126049786475e-06,
+      "loss": 0.3462,
+      "step": 20262
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3292894700188206e-06,
+      "loss": 0.3087,
+      "step": 20263
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328366406283439e-06,
+      "loss": 0.3326,
+      "step": 20264
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.327443413784089e-06,
+      "loss": 0.3497,
+      "step": 20265
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3265204925323714e-06,
+      "loss": 0.3241,
+      "step": 20266
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.32559764253987e-06,
+      "loss": 0.3261,
+      "step": 20267
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.324674863818185e-06,
+      "loss": 0.3283,
+      "step": 20268
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.323752156378896e-06,
+      "loss": 0.2966,
+      "step": 20269
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322829520233599e-06,
+      "loss": 0.2899,
+      "step": 20270
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3219069553938856e-06,
+      "loss": 0.3084,
+      "step": 20271
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3209844618713335e-06,
+      "loss": 0.3193,
+      "step": 20272
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320062039677536e-06,
+      "loss": 0.317,
+      "step": 20273
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.31913968882408e-06,
+      "loss": 0.3305,
+      "step": 20274
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318217409322545e-06,
+      "loss": 0.3247,
+      "step": 20275
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.31729520118452e-06,
+      "loss": 0.3125,
+      "step": 20276
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3163730644215865e-06,
+      "loss": 0.2842,
+      "step": 20277
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315450999045322e-06,
+      "loss": 0.3373,
+      "step": 20278
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.31452900506731e-06,
+      "loss": 0.309,
+      "step": 20279
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.313607082499137e-06,
+      "loss": 0.3199,
+      "step": 20280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.312685231352374e-06,
+      "loss": 0.3277,
+      "step": 20281
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.311763451638603e-06,
+      "loss": 0.3277,
+      "step": 20282
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310841743369405e-06,
+      "loss": 0.3223,
+      "step": 20283
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.309920106556349e-06,
+      "loss": 0.3198,
+      "step": 20284
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308998541211016e-06,
+      "loss": 0.3317,
+      "step": 20285
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308077047344984e-06,
+      "loss": 0.3346,
+      "step": 20286
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.307155624969822e-06,
+      "loss": 0.3154,
+      "step": 20287
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3062342740971004e-06,
+      "loss": 0.3355,
+      "step": 20288
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3053129947384006e-06,
+      "loss": 0.2958,
+      "step": 20289
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304391786905283e-06,
+      "loss": 0.3213,
+      "step": 20290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.303470650609325e-06,
+      "loss": 0.3082,
+      "step": 20291
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.302549585862098e-06,
+      "loss": 0.3326,
+      "step": 20292
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.301628592675163e-06,
+      "loss": 0.2916,
+      "step": 20293
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3007076710600925e-06,
+      "loss": 0.3421,
+      "step": 20294
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299786821028456e-06,
+      "loss": 0.2949,
+      "step": 20295
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2988660425918125e-06,
+      "loss": 0.3446,
+      "step": 20296
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297945335761735e-06,
+      "loss": 0.3229,
+      "step": 20297
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297024700549784e-06,
+      "loss": 0.2852,
+      "step": 20298
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.296104136967518e-06,
+      "loss": 0.3153,
+      "step": 20299
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295183645026503e-06,
+      "loss": 0.3086,
+      "step": 20300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.294263224738305e-06,
+      "loss": 0.3151,
+      "step": 20301
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293342876114477e-06,
+      "loss": 0.3256,
+      "step": 20302
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292422599166581e-06,
+      "loss": 0.3095,
+      "step": 20303
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.291502393906179e-06,
+      "loss": 0.3547,
+      "step": 20304
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290582260344828e-06,
+      "loss": 0.3507,
+      "step": 20305
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.289662198494081e-06,
+      "loss": 0.3616,
+      "step": 20306
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2887422083655006e-06,
+      "loss": 0.3433,
+      "step": 20307
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.287822289970637e-06,
+      "loss": 0.2717,
+      "step": 20308
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2869024433210425e-06,
+      "loss": 0.3285,
+      "step": 20309
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285982668428272e-06,
+      "loss": 0.327,
+      "step": 20310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285062965303884e-06,
+      "loss": 0.3098,
+      "step": 20311
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.284143333959422e-06,
+      "loss": 0.2839,
+      "step": 20312
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2832237744064386e-06,
+      "loss": 0.2799,
+      "step": 20313
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282304286656489e-06,
+      "loss": 0.2987,
+      "step": 20314
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2813848707211145e-06,
+      "loss": 0.3243,
+      "step": 20315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.280465526611866e-06,
+      "loss": 0.3257,
+      "step": 20316
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279546254340294e-06,
+      "loss": 0.3355,
+      "step": 20317
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278627053917942e-06,
+      "loss": 0.3119,
+      "step": 20318
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.277707925356351e-06,
+      "loss": 0.3185,
+      "step": 20319
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2767888686670725e-06,
+      "loss": 0.2928,
+      "step": 20320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.275869883861643e-06,
+      "loss": 0.3226,
+      "step": 20321
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274950970951608e-06,
+      "loss": 0.3236,
+      "step": 20322
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274032129948512e-06,
+      "loss": 0.3217,
+      "step": 20323
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2731133608638906e-06,
+      "loss": 0.352,
+      "step": 20324
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.272194663709287e-06,
+      "loss": 0.3504,
+      "step": 20325
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271276038496242e-06,
+      "loss": 0.3312,
+      "step": 20326
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.270357485236287e-06,
+      "loss": 0.3102,
+      "step": 20327
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269439003940966e-06,
+      "loss": 0.347,
+      "step": 20328
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268520594621809e-06,
+      "loss": 0.3188,
+      "step": 20329
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267602257290358e-06,
+      "loss": 0.305,
+      "step": 20330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26668399195814e-06,
+      "loss": 0.3014,
+      "step": 20331
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2657657986366966e-06,
+      "loss": 0.3278,
+      "step": 20332
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.264847677337551e-06,
+      "loss": 0.3001,
+      "step": 20333
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263929628072241e-06,
+      "loss": 0.3352,
+      "step": 20334
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263011650852299e-06,
+      "loss": 0.2962,
+      "step": 20335
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.262093745689248e-06,
+      "loss": 0.3292,
+      "step": 20336
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.261175912594622e-06,
+      "loss": 0.3216,
+      "step": 20337
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26025815157995e-06,
+      "loss": 0.335,
+      "step": 20338
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.259340462656755e-06,
+      "loss": 0.289,
+      "step": 20339
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258422845836568e-06,
+      "loss": 0.335,
+      "step": 20340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.257505301130909e-06,
+      "loss": 0.332,
+      "step": 20341
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256587828551307e-06,
+      "loss": 0.3388,
+      "step": 20342
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255670428109281e-06,
+      "loss": 0.2816,
+      "step": 20343
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2547530998163565e-06,
+      "loss": 0.306,
+      "step": 20344
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253835843684057e-06,
+      "loss": 0.2919,
+      "step": 20345
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.252918659723899e-06,
+      "loss": 0.3255,
+      "step": 20346
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2520015479474046e-06,
+      "loss": 0.3122,
+      "step": 20347
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.251084508366095e-06,
+      "loss": 0.3121,
+      "step": 20348
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250167540991483e-06,
+      "loss": 0.3248,
+      "step": 20349
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.249250645835092e-06,
+      "loss": 0.3348,
+      "step": 20350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.248333822908436e-06,
+      "loss": 0.3001,
+      "step": 20351
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247417072223025e-06,
+      "loss": 0.3159,
+      "step": 20352
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246500393790378e-06,
+      "loss": 0.3239,
+      "step": 20353
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.245583787622012e-06,
+      "loss": 0.3434,
+      "step": 20354
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244667253729431e-06,
+      "loss": 0.2983,
+      "step": 20355
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243750792124153e-06,
+      "loss": 0.3223,
+      "step": 20356
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24283440281769e-06,
+      "loss": 0.3257,
+      "step": 20357
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.241918085821547e-06,
+      "loss": 0.2935,
+      "step": 20358
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.241001841147235e-06,
+      "loss": 0.3203,
+      "step": 20359
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240085668806266e-06,
+      "loss": 0.3305,
+      "step": 20360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.239169568810143e-06,
+      "loss": 0.3468,
+      "step": 20361
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238253541170369e-06,
+      "loss": 0.3385,
+      "step": 20362
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2373375858984575e-06,
+      "loss": 0.3028,
+      "step": 20363
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.236421703005905e-06,
+      "loss": 0.2965,
+      "step": 20364
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.235505892504219e-06,
+      "loss": 0.3538,
+      "step": 20365
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234590154404905e-06,
+      "loss": 0.3308,
+      "step": 20366
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233674488719458e-06,
+      "loss": 0.3157,
+      "step": 20367
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232758895459382e-06,
+      "loss": 0.3102,
+      "step": 20368
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.231843374636182e-06,
+      "loss": 0.3225,
+      "step": 20369
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230927926261348e-06,
+      "loss": 0.2557,
+      "step": 20370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230012550346385e-06,
+      "loss": 0.3066,
+      "step": 20371
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.229097246902789e-06,
+      "loss": 0.2859,
+      "step": 20372
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.228182015942051e-06,
+      "loss": 0.3313,
+      "step": 20373
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227266857475669e-06,
+      "loss": 0.3007,
+      "step": 20374
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.226351771515144e-06,
+      "loss": 0.3374,
+      "step": 20375
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.225436758071959e-06,
+      "loss": 0.3379,
+      "step": 20376
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.224521817157612e-06,
+      "loss": 0.3218,
+      "step": 20377
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223606948783594e-06,
+      "loss": 0.3403,
+      "step": 20378
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222692152961401e-06,
+      "loss": 0.325,
+      "step": 20379
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221777429702514e-06,
+      "loss": 0.3339,
+      "step": 20380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.22086277901843e-06,
+      "loss": 0.3295,
+      "step": 20381
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219948200920633e-06,
+      "loss": 0.3522,
+      "step": 20382
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219033695420606e-06,
+      "loss": 0.3007,
+      "step": 20383
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2181192625298395e-06,
+      "loss": 0.3506,
+      "step": 20384
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217204902259824e-06,
+      "loss": 0.3125,
+      "step": 20385
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216290614622034e-06,
+      "loss": 0.3325,
+      "step": 20386
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.215376399627959e-06,
+      "loss": 0.3058,
+      "step": 20387
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.214462257289083e-06,
+      "loss": 0.3307,
+      "step": 20388
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2135481876168805e-06,
+      "loss": 0.3384,
+      "step": 20389
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.212634190622838e-06,
+      "loss": 0.2853,
+      "step": 20390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211720266318437e-06,
+      "loss": 0.3284,
+      "step": 20391
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210806414715155e-06,
+      "loss": 0.3052,
+      "step": 20392
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209892635824464e-06,
+      "loss": 0.3473,
+      "step": 20393
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20897892965785e-06,
+      "loss": 0.2767,
+      "step": 20394
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208065296226781e-06,
+      "loss": 0.3001,
+      "step": 20395
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.207151735542737e-06,
+      "loss": 0.3544,
+      "step": 20396
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206238247617195e-06,
+      "loss": 0.324,
+      "step": 20397
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20532483246162e-06,
+      "loss": 0.3199,
+      "step": 20398
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204411490087491e-06,
+      "loss": 0.2986,
+      "step": 20399
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.203498220506282e-06,
+      "loss": 0.314,
+      "step": 20400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202585023729456e-06,
+      "loss": 0.3075,
+      "step": 20401
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20167189976849e-06,
+      "loss": 0.3168,
+      "step": 20402
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200758848634851e-06,
+      "loss": 0.3081,
+      "step": 20403
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.199845870340001e-06,
+      "loss": 0.3123,
+      "step": 20404
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198932964895412e-06,
+      "loss": 0.3405,
+      "step": 20405
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1980201323125535e-06,
+      "loss": 0.3213,
+      "step": 20406
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197107372602883e-06,
+      "loss": 0.3193,
+      "step": 20407
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196194685777869e-06,
+      "loss": 0.3269,
+      "step": 20408
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19528207184898e-06,
+      "loss": 0.3292,
+      "step": 20409
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1943695308276675e-06,
+      "loss": 0.3015,
+      "step": 20410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193457062725399e-06,
+      "loss": 0.303,
+      "step": 20411
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19254466755364e-06,
+      "loss": 0.3147,
+      "step": 20412
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.191632345323844e-06,
+      "loss": 0.3012,
+      "step": 20413
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190720096047468e-06,
+      "loss": 0.3167,
+      "step": 20414
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189807919735973e-06,
+      "loss": 0.3272,
+      "step": 20415
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18889581640082e-06,
+      "loss": 0.3266,
+      "step": 20416
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187983786053457e-06,
+      "loss": 0.317,
+      "step": 20417
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187071828705342e-06,
+      "loss": 0.2768,
+      "step": 20418
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186159944367936e-06,
+      "loss": 0.2992,
+      "step": 20419
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185248133052681e-06,
+      "loss": 0.3377,
+      "step": 20420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.184336394771037e-06,
+      "loss": 0.3137,
+      "step": 20421
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183424729534456e-06,
+      "loss": 0.315,
+      "step": 20422
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.182513137354387e-06,
+      "loss": 0.2661,
+      "step": 20423
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181601618242275e-06,
+      "loss": 0.3371,
+      "step": 20424
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1806901722095746e-06,
+      "loss": 0.3351,
+      "step": 20425
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.179778799267729e-06,
+      "loss": 0.3123,
+      "step": 20426
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1788674994281886e-06,
+      "loss": 0.2801,
+      "step": 20427
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1779562727024015e-06,
+      "loss": 0.3394,
+      "step": 20428
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177045119101806e-06,
+      "loss": 0.3357,
+      "step": 20429
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17613403863785e-06,
+      "loss": 0.3075,
+      "step": 20430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17522303132198e-06,
+      "loss": 0.3185,
+      "step": 20431
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1743120971656315e-06,
+      "loss": 0.3316,
+      "step": 20432
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173401236180252e-06,
+      "loss": 0.2915,
+      "step": 20433
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172490448377277e-06,
+      "loss": 0.3163,
+      "step": 20434
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.171579733768152e-06,
+      "loss": 0.3081,
+      "step": 20435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170669092364308e-06,
+      "loss": 0.3344,
+      "step": 20436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.16975852417719e-06,
+      "loss": 0.3537,
+      "step": 20437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168848029218227e-06,
+      "loss": 0.3269,
+      "step": 20438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.167937607498861e-06,
+      "loss": 0.2552,
+      "step": 20439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.167027259030528e-06,
+      "loss": 0.3171,
+      "step": 20440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166116983824656e-06,
+      "loss": 0.3234,
+      "step": 20441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165206781892682e-06,
+      "loss": 0.3126,
+      "step": 20442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164296653246041e-06,
+      "loss": 0.297,
+      "step": 20443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163386597896156e-06,
+      "loss": 0.328,
+      "step": 20444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.162476615854467e-06,
+      "loss": 0.3379,
+      "step": 20445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161566707132399e-06,
+      "loss": 0.3435,
+      "step": 20446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1606568717413755e-06,
+      "loss": 0.3139,
+      "step": 20447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.159747109692829e-06,
+      "loss": 0.3312,
+      "step": 20448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15883742099819e-06,
+      "loss": 0.3219,
+      "step": 20449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.157927805668875e-06,
+      "loss": 0.3142,
+      "step": 20450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1570182637163155e-06,
+      "loss": 0.3174,
+      "step": 20451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156108795151933e-06,
+      "loss": 0.3256,
+      "step": 20452
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155199399987156e-06,
+      "loss": 0.3224,
+      "step": 20453
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154290078233397e-06,
+      "loss": 0.3057,
+      "step": 20454
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153380829902087e-06,
+      "loss": 0.3025,
+      "step": 20455
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152471655004641e-06,
+      "loss": 0.3147,
+      "step": 20456
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.151562553552474e-06,
+      "loss": 0.3037,
+      "step": 20457
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.150653525557009e-06,
+      "loss": 0.3345,
+      "step": 20458
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149744571029667e-06,
+      "loss": 0.2776,
+      "step": 20459
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.148835689981857e-06,
+      "loss": 0.3279,
+      "step": 20460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147926882424999e-06,
+      "loss": 0.3152,
+      "step": 20461
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1470181483705105e-06,
+      "loss": 0.335,
+      "step": 20462
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1461094878297994e-06,
+      "loss": 0.3117,
+      "step": 20463
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.145200900814279e-06,
+      "loss": 0.2975,
+      "step": 20464
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144292387335368e-06,
+      "loss": 0.3206,
+      "step": 20465
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.143383947404472e-06,
+      "loss": 0.3278,
+      "step": 20466
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142475581032999e-06,
+      "loss": 0.2968,
+      "step": 20467
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.141567288232363e-06,
+      "loss": 0.3203,
+      "step": 20468
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.140659069013967e-06,
+      "loss": 0.3344,
+      "step": 20469
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1397509233892215e-06,
+      "loss": 0.309,
+      "step": 20470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.138842851369536e-06,
+      "loss": 0.3398,
+      "step": 20471
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.137934852966309e-06,
+      "loss": 0.3233,
+      "step": 20472
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.137026928190948e-06,
+      "loss": 0.3159,
+      "step": 20473
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136119077054862e-06,
+      "loss": 0.3142,
+      "step": 20474
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1352112995694436e-06,
+      "loss": 0.3475,
+      "step": 20475
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.134303595746104e-06,
+      "loss": 0.3556,
+      "step": 20476
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.133395965596241e-06,
+      "loss": 0.3386,
+      "step": 20477
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132488409131248e-06,
+      "loss": 0.3053,
+      "step": 20478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.131580926362529e-06,
+      "loss": 0.3108,
+      "step": 20479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1306735173014875e-06,
+      "loss": 0.3104,
+      "step": 20480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.129766181959511e-06,
+      "loss": 0.3487,
+      "step": 20481
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.128858920348e-06,
+      "loss": 0.3466,
+      "step": 20482
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127951732478353e-06,
+      "loss": 0.3114,
+      "step": 20483
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127044618361959e-06,
+      "loss": 0.3395,
+      "step": 20484
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.126137578010211e-06,
+      "loss": 0.3115,
+      "step": 20485
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1252306114345085e-06,
+      "loss": 0.3245,
+      "step": 20486
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124323718646239e-06,
+      "loss": 0.3455,
+      "step": 20487
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.123416899656789e-06,
+      "loss": 0.338,
+      "step": 20488
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.122510154477552e-06,
+      "loss": 0.3345,
+      "step": 20489
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12160348311992e-06,
+      "loss": 0.3398,
+      "step": 20490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.120696885595274e-06,
+      "loss": 0.311,
+      "step": 20491
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119790361915005e-06,
+      "loss": 0.3293,
+      "step": 20492
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.118883912090501e-06,
+      "loss": 0.293,
+      "step": 20493
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117977536133142e-06,
+      "loss": 0.2881,
+      "step": 20494
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117071234054313e-06,
+      "loss": 0.3229,
+      "step": 20495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.116165005865403e-06,
+      "loss": 0.2932,
+      "step": 20496
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11525885157779e-06,
+      "loss": 0.3307,
+      "step": 20497
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114352771202853e-06,
+      "loss": 0.3108,
+      "step": 20498
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113446764751977e-06,
+      "loss": 0.3676,
+      "step": 20499
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112540832236536e-06,
+      "loss": 0.3285,
+      "step": 20500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111634973667912e-06,
+      "loss": 0.2946,
+      "step": 20501
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110729189057487e-06,
+      "loss": 0.3181,
+      "step": 20502
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109823478416628e-06,
+      "loss": 0.3088,
+      "step": 20503
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108917841756717e-06,
+      "loss": 0.3101,
+      "step": 20504
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108012279089131e-06,
+      "loss": 0.3395,
+      "step": 20505
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.107106790425238e-06,
+      "loss": 0.3056,
+      "step": 20506
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.106201375776416e-06,
+      "loss": 0.3024,
+      "step": 20507
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1052960351540325e-06,
+      "loss": 0.3378,
+      "step": 20508
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104390768569464e-06,
+      "loss": 0.2959,
+      "step": 20509
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103485576034074e-06,
+      "loss": 0.3278,
+      "step": 20510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.102580457559238e-06,
+      "loss": 0.3097,
+      "step": 20511
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10167541315632e-06,
+      "loss": 0.2958,
+      "step": 20512
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100770442836688e-06,
+      "loss": 0.3004,
+      "step": 20513
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099865546611714e-06,
+      "loss": 0.3212,
+      "step": 20514
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098960724492756e-06,
+      "loss": 0.2993,
+      "step": 20515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.09805597649118e-06,
+      "loss": 0.3237,
+      "step": 20516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.097151302618356e-06,
+      "loss": 0.3173,
+      "step": 20517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096246702885639e-06,
+      "loss": 0.3253,
+      "step": 20518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095342177304397e-06,
+      "loss": 0.3456,
+      "step": 20519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.094437725885988e-06,
+      "loss": 0.3141,
+      "step": 20520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093533348641769e-06,
+      "loss": 0.2823,
+      "step": 20521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0926290455831015e-06,
+      "loss": 0.294,
+      "step": 20522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091724816721344e-06,
+      "loss": 0.2978,
+      "step": 20523
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090820662067857e-06,
+      "loss": 0.3106,
+      "step": 20524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08991658163399e-06,
+      "loss": 0.3027,
+      "step": 20525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.089012575431103e-06,
+      "loss": 0.3277,
+      "step": 20526
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0881086434705505e-06,
+      "loss": 0.3339,
+      "step": 20527
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0872047857636835e-06,
+      "loss": 0.3243,
+      "step": 20528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0863010023218575e-06,
+      "loss": 0.3546,
+      "step": 20529
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0853972931564225e-06,
+      "loss": 0.3054,
+      "step": 20530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.084493658278724e-06,
+      "loss": 0.3251,
+      "step": 20531
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.083590097700118e-06,
+      "loss": 0.3131,
+      "step": 20532
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082686611431954e-06,
+      "loss": 0.3202,
+      "step": 20533
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081783199485575e-06,
+      "loss": 0.3339,
+      "step": 20534
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08087986187233e-06,
+      "loss": 0.3126,
+      "step": 20535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07997659860357e-06,
+      "loss": 0.3275,
+      "step": 20536
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07907340969063e-06,
+      "loss": 0.3543,
+      "step": 20537
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078170295144862e-06,
+      "loss": 0.2998,
+      "step": 20538
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077267254977608e-06,
+      "loss": 0.2838,
+      "step": 20539
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076364289200211e-06,
+      "loss": 0.298,
+      "step": 20540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.075461397824005e-06,
+      "loss": 0.3298,
+      "step": 20541
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074558580860341e-06,
+      "loss": 0.3046,
+      "step": 20542
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.073655838320548e-06,
+      "loss": 0.2986,
+      "step": 20543
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.072753170215969e-06,
+      "loss": 0.3389,
+      "step": 20544
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071850576557948e-06,
+      "loss": 0.3395,
+      "step": 20545
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070948057357811e-06,
+      "loss": 0.2859,
+      "step": 20546
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070045612626898e-06,
+      "loss": 0.2744,
+      "step": 20547
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.069143242376549e-06,
+      "loss": 0.3544,
+      "step": 20548
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068240946618088e-06,
+      "loss": 0.3146,
+      "step": 20549
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067338725362857e-06,
+      "loss": 0.2943,
+      "step": 20550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066436578622185e-06,
+      "loss": 0.2994,
+      "step": 20551
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065534506407397e-06,
+      "loss": 0.3038,
+      "step": 20552
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.064632508729829e-06,
+      "loss": 0.3466,
+      "step": 20553
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0637305856008115e-06,
+      "loss": 0.3283,
+      "step": 20554
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062828737031668e-06,
+      "loss": 0.3454,
+      "step": 20555
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.061926963033727e-06,
+      "loss": 0.3312,
+      "step": 20556
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.061025263618319e-06,
+      "loss": 0.2954,
+      "step": 20557
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060123638796763e-06,
+      "loss": 0.3079,
+      "step": 20558
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0592220885803875e-06,
+      "loss": 0.3109,
+      "step": 20559
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05832061298052e-06,
+      "loss": 0.3309,
+      "step": 20560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057419212008476e-06,
+      "loss": 0.3171,
+      "step": 20561
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.056517885675579e-06,
+      "loss": 0.3235,
+      "step": 20562
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.055616633993147e-06,
+      "loss": 0.292,
+      "step": 20563
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054715456972508e-06,
+      "loss": 0.3211,
+      "step": 20564
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.053814354624972e-06,
+      "loss": 0.3313,
+      "step": 20565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05291332696186e-06,
+      "loss": 0.3047,
+      "step": 20566
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.052012373994495e-06,
+      "loss": 0.2879,
+      "step": 20567
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0511114957341835e-06,
+      "loss": 0.3069,
+      "step": 20568
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050210692192244e-06,
+      "loss": 0.3098,
+      "step": 20569
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.049309963379996e-06,
+      "loss": 0.3257,
+      "step": 20570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.048409309308748e-06,
+      "loss": 0.3179,
+      "step": 20571
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0475087299898076e-06,
+      "loss": 0.2935,
+      "step": 20572
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.046608225434497e-06,
+      "loss": 0.3121,
+      "step": 20573
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045707795654113e-06,
+      "loss": 0.3374,
+      "step": 20574
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0448074406599756e-06,
+      "loss": 0.296,
+      "step": 20575
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043907160463394e-06,
+      "loss": 0.3402,
+      "step": 20576
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043006955075667e-06,
+      "loss": 0.2948,
+      "step": 20577
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.042106824508107e-06,
+      "loss": 0.3134,
+      "step": 20578
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.041206768772023e-06,
+      "loss": 0.3149,
+      "step": 20579
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0403067878787105e-06,
+      "loss": 0.2909,
+      "step": 20580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.039406881839484e-06,
+      "loss": 0.286,
+      "step": 20581
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0385070506656395e-06,
+      "loss": 0.3261,
+      "step": 20582
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.037607294368478e-06,
+      "loss": 0.3154,
+      "step": 20583
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.036707612959302e-06,
+      "loss": 0.3205,
+      "step": 20584
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.035808006449415e-06,
+      "loss": 0.3046,
+      "step": 20585
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034908474850112e-06,
+      "loss": 0.2954,
+      "step": 20586
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034009018172691e-06,
+      "loss": 0.2896,
+      "step": 20587
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033109636428455e-06,
+      "loss": 0.2562,
+      "step": 20588
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.032210329628693e-06,
+      "loss": 0.2927,
+      "step": 20589
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031311097784703e-06,
+      "loss": 0.3172,
+      "step": 20590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.030411940907784e-06,
+      "loss": 0.3114,
+      "step": 20591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0295128590092245e-06,
+      "loss": 0.3127,
+      "step": 20592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028613852100315e-06,
+      "loss": 0.2858,
+      "step": 20593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.027714920192353e-06,
+      "loss": 0.2974,
+      "step": 20594
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026816063296622e-06,
+      "loss": 0.3364,
+      "step": 20595
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025917281424417e-06,
+      "loss": 0.2991,
+      "step": 20596
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0250185745870254e-06,
+      "loss": 0.3125,
+      "step": 20597
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.024119942795739e-06,
+      "loss": 0.3114,
+      "step": 20598
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023221386061836e-06,
+      "loss": 0.3154,
+      "step": 20599
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0223229043966084e-06,
+      "loss": 0.3187,
+      "step": 20600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.021424497811343e-06,
+      "loss": 0.3254,
+      "step": 20601
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.02052616631732e-06,
+      "loss": 0.3585,
+      "step": 20602
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0196279099258205e-06,
+      "loss": 0.2845,
+      "step": 20603
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.018729728648132e-06,
+      "loss": 0.3189,
+      "step": 20604
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017831622495531e-06,
+      "loss": 0.3404,
+      "step": 20605
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.016933591479299e-06,
+      "loss": 0.321,
+      "step": 20606
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01603563561072e-06,
+      "loss": 0.2826,
+      "step": 20607
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.015137754901064e-06,
+      "loss": 0.3096,
+      "step": 20608
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014239949361615e-06,
+      "loss": 0.3145,
+      "step": 20609
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.013342219003648e-06,
+      "loss": 0.3365,
+      "step": 20610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012444563838437e-06,
+      "loss": 0.3151,
+      "step": 20611
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011546983877261e-06,
+      "loss": 0.2921,
+      "step": 20612
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0106494791313865e-06,
+      "loss": 0.3555,
+      "step": 20613
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009752049612092e-06,
+      "loss": 0.3159,
+      "step": 20614
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0088546953306464e-06,
+      "loss": 0.3259,
+      "step": 20615
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0079574162983226e-06,
+      "loss": 0.3679,
+      "step": 20616
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007060212526387e-06,
+      "loss": 0.3088,
+      "step": 20617
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00616308402611e-06,
+      "loss": 0.2886,
+      "step": 20618
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.005266030808764e-06,
+      "loss": 0.3026,
+      "step": 20619
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0043690528856095e-06,
+      "loss": 0.3031,
+      "step": 20620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0034721502679155e-06,
+      "loss": 0.3247,
+      "step": 20621
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00257532296695e-06,
+      "loss": 0.3322,
+      "step": 20622
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001678570993971e-06,
+      "loss": 0.3232,
+      "step": 20623
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000781894360249e-06,
+      "loss": 0.3144,
+      "step": 20624
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999885293077041e-06,
+      "loss": 0.3084,
+      "step": 20625
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998988767155608e-06,
+      "loss": 0.341,
+      "step": 20626
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998092316607212e-06,
+      "loss": 0.3218,
+      "step": 20627
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997195941443115e-06,
+      "loss": 0.3194,
+      "step": 20628
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99629964167457e-06,
+      "loss": 0.3165,
+      "step": 20629
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995403417312839e-06,
+      "loss": 0.2852,
+      "step": 20630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99450726836918e-06,
+      "loss": 0.3304,
+      "step": 20631
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.993611194854842e-06,
+      "loss": 0.314,
+      "step": 20632
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.992715196781084e-06,
+      "loss": 0.3303,
+      "step": 20633
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.991819274159163e-06,
+      "loss": 0.3472,
+      "step": 20634
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990923427000329e-06,
+      "loss": 0.3203,
+      "step": 20635
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99002765531583e-06,
+      "loss": 0.3203,
+      "step": 20636
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989131959116919e-06,
+      "loss": 0.3125,
+      "step": 20637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.988236338414853e-06,
+      "loss": 0.3666,
+      "step": 20638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98734079322087e-06,
+      "loss": 0.3146,
+      "step": 20639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9864453235462234e-06,
+      "loss": 0.3323,
+      "step": 20640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985549929402164e-06,
+      "loss": 0.3119,
+      "step": 20641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.984654610799931e-06,
+      "loss": 0.3165,
+      "step": 20642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983759367750772e-06,
+      "loss": 0.3047,
+      "step": 20643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982864200265938e-06,
+      "loss": 0.3473,
+      "step": 20644
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981969108356665e-06,
+      "loss": 0.3067,
+      "step": 20645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981074092034194e-06,
+      "loss": 0.3305,
+      "step": 20646
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.980179151309773e-06,
+      "loss": 0.3296,
+      "step": 20647
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979284286194636e-06,
+      "loss": 0.3381,
+      "step": 20648
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.978389496700024e-06,
+      "loss": 0.2786,
+      "step": 20649
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.977494782837182e-06,
+      "loss": 0.3305,
+      "step": 20650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9766001446173395e-06,
+      "loss": 0.2927,
+      "step": 20651
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975705582051735e-06,
+      "loss": 0.3108,
+      "step": 20652
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97481109515161e-06,
+      "loss": 0.3205,
+      "step": 20653
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973916683928193e-06,
+      "loss": 0.3497,
+      "step": 20654
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973022348392722e-06,
+      "loss": 0.3367,
+      "step": 20655
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9721280885564285e-06,
+      "loss": 0.3301,
+      "step": 20656
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9712339044305404e-06,
+      "loss": 0.2894,
+      "step": 20657
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970339796026291e-06,
+      "loss": 0.3025,
+      "step": 20658
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.969445763354917e-06,
+      "loss": 0.2852,
+      "step": 20659
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968551806427636e-06,
+      "loss": 0.3384,
+      "step": 20660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9676579252556835e-06,
+      "loss": 0.3072,
+      "step": 20661
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96676411985029e-06,
+      "loss": 0.3052,
+      "step": 20662
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965870390222671e-06,
+      "loss": 0.3001,
+      "step": 20663
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.964976736384059e-06,
+      "loss": 0.3208,
+      "step": 20664
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.964083158345681e-06,
+      "loss": 0.2881,
+      "step": 20665
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.963189656118756e-06,
+      "loss": 0.3274,
+      "step": 20666
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962296229714504e-06,
+      "loss": 0.299,
+      "step": 20667
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.961402879144153e-06,
+      "loss": 0.3066,
+      "step": 20668
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.960509604418916e-06,
+      "loss": 0.3542,
+      "step": 20669
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959616405550016e-06,
+      "loss": 0.3102,
+      "step": 20670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9587232825486735e-06,
+      "loss": 0.3324,
+      "step": 20671
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9578302354261085e-06,
+      "loss": 0.3056,
+      "step": 20672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95693726419353e-06,
+      "loss": 0.3157,
+      "step": 20673
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.956044368862158e-06,
+      "loss": 0.3219,
+      "step": 20674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955151549443211e-06,
+      "loss": 0.2886,
+      "step": 20675
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954258805947899e-06,
+      "loss": 0.3301,
+      "step": 20676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953366138387431e-06,
+      "loss": 0.258,
+      "step": 20677
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.952473546773028e-06,
+      "loss": 0.2826,
+      "step": 20678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951581031115891e-06,
+      "loss": 0.3254,
+      "step": 20679
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.950688591427235e-06,
+      "loss": 0.3,
+      "step": 20680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949796227718274e-06,
+      "loss": 0.3144,
+      "step": 20681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948903940000206e-06,
+      "loss": 0.3097,
+      "step": 20682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948011728284244e-06,
+      "loss": 0.3393,
+      "step": 20683
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947119592581597e-06,
+      "loss": 0.3168,
+      "step": 20684
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.946227532903464e-06,
+      "loss": 0.3058,
+      "step": 20685
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945335549261056e-06,
+      "loss": 0.3046,
+      "step": 20686
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9444436416655715e-06,
+      "loss": 0.3146,
+      "step": 20687
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943551810128211e-06,
+      "loss": 0.3292,
+      "step": 20688
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.942660054660179e-06,
+      "loss": 0.3231,
+      "step": 20689
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.941768375272679e-06,
+      "loss": 0.3436,
+      "step": 20690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940876771976905e-06,
+      "loss": 0.3096,
+      "step": 20691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939985244784057e-06,
+      "loss": 0.3064,
+      "step": 20692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939093793705338e-06,
+      "loss": 0.3112,
+      "step": 20693
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9382024187519365e-06,
+      "loss": 0.3183,
+      "step": 20694
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937311119935052e-06,
+      "loss": 0.3207,
+      "step": 20695
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9364198972658826e-06,
+      "loss": 0.3253,
+      "step": 20696
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9355287507556205e-06,
+      "loss": 0.3404,
+      "step": 20697
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934637680415452e-06,
+      "loss": 0.3327,
+      "step": 20698
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933746686256578e-06,
+      "loss": 0.3055,
+      "step": 20699
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932855768290181e-06,
+      "loss": 0.3273,
+      "step": 20700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9319649265274566e-06,
+      "loss": 0.2822,
+      "step": 20701
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931074160979595e-06,
+      "loss": 0.2925,
+      "step": 20702
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9301834716577795e-06,
+      "loss": 0.2926,
+      "step": 20703
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929292858573199e-06,
+      "loss": 0.2925,
+      "step": 20704
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92840232173704e-06,
+      "loss": 0.3021,
+      "step": 20705
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927511861160491e-06,
+      "loss": 0.336,
+      "step": 20706
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9266214768547335e-06,
+      "loss": 0.2884,
+      "step": 20707
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925731168830948e-06,
+      "loss": 0.3314,
+      "step": 20708
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9248409371003216e-06,
+      "loss": 0.2978,
+      "step": 20709
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92395078167403e-06,
+      "loss": 0.3257,
+      "step": 20710
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923060702563256e-06,
+      "loss": 0.3616,
+      "step": 20711
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922170699779184e-06,
+      "loss": 0.2943,
+      "step": 20712
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921280773332985e-06,
+      "loss": 0.329,
+      "step": 20713
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92039092323584e-06,
+      "loss": 0.301,
+      "step": 20714
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.91950114949893e-06,
+      "loss": 0.3368,
+      "step": 20715
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918611452133421e-06,
+      "loss": 0.3507,
+      "step": 20716
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917721831150497e-06,
+      "loss": 0.3342,
+      "step": 20717
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.916832286561324e-06,
+      "loss": 0.2945,
+      "step": 20718
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915942818377081e-06,
+      "loss": 0.3395,
+      "step": 20719
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915053426608935e-06,
+      "loss": 0.3434,
+      "step": 20720
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.914164111268061e-06,
+      "loss": 0.3573,
+      "step": 20721
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913274872365622e-06,
+      "loss": 0.3375,
+      "step": 20722
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 0.327,
+      "step": 20723
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911496623920744e-06,
+      "loss": 0.3327,
+      "step": 20724
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.910607614400634e-06,
+      "loss": 0.3552,
+      "step": 20725
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.909718681363634e-06,
+      "loss": 0.3237,
+      "step": 20726
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908829824820913e-06,
+      "loss": 0.341,
+      "step": 20727
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907941044783626e-06,
+      "loss": 0.318,
+      "step": 20728
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9070523412629445e-06,
+      "loss": 0.3253,
+      "step": 20729
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906163714270026e-06,
+      "loss": 0.3024,
+      "step": 20730
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90527516381603e-06,
+      "loss": 0.3112,
+      "step": 20731
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.904386689912119e-06,
+      "loss": 0.3001,
+      "step": 20732
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903498292569455e-06,
+      "loss": 0.2853,
+      "step": 20733
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902609971799191e-06,
+      "loss": 0.3633,
+      "step": 20734
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901721727612489e-06,
+      "loss": 0.3154,
+      "step": 20735
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900833560020505e-06,
+      "loss": 0.3557,
+      "step": 20736
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899945469034391e-06,
+      "loss": 0.3507,
+      "step": 20737
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8990574546653026e-06,
+      "loss": 0.3343,
+      "step": 20738
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8981695169243986e-06,
+      "loss": 0.2956,
+      "step": 20739
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.897281655822826e-06,
+      "loss": 0.3283,
+      "step": 20740
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.896393871371734e-06,
+      "loss": 0.3103,
+      "step": 20741
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895506163582277e-06,
+      "loss": 0.3281,
+      "step": 20742
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89461853246561e-06,
+      "loss": 0.353,
+      "step": 20743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89373097803287e-06,
+      "loss": 0.3199,
+      "step": 20744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892843500295212e-06,
+      "loss": 0.31,
+      "step": 20745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891956099263786e-06,
+      "loss": 0.3161,
+      "step": 20746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891068774949729e-06,
+      "loss": 0.3247,
+      "step": 20747
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89018152736419e-06,
+      "loss": 0.2831,
+      "step": 20748
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8892943565183186e-06,
+      "loss": 0.3116,
+      "step": 20749
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888407262423251e-06,
+      "loss": 0.2935,
+      "step": 20750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887520245090126e-06,
+      "loss": 0.3491,
+      "step": 20751
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.886633304530094e-06,
+      "loss": 0.3084,
+      "step": 20752
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.885746440754285e-06,
+      "loss": 0.3116,
+      "step": 20753
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.884859653773845e-06,
+      "loss": 0.2958,
+      "step": 20754
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883972943599913e-06,
+      "loss": 0.3047,
+      "step": 20755
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88308631024362e-06,
+      "loss": 0.3028,
+      "step": 20756
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882199753716105e-06,
+      "loss": 0.2903,
+      "step": 20757
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.881313274028508e-06,
+      "loss": 0.3298,
+      "step": 20758
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8804268711919545e-06,
+      "loss": 0.323,
+      "step": 20759
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.879540545217587e-06,
+      "loss": 0.323,
+      "step": 20760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.878654296116532e-06,
+      "loss": 0.3269,
+      "step": 20761
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877768123899918e-06,
+      "loss": 0.3331,
+      "step": 20762
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.876882028578881e-06,
+      "loss": 0.3277,
+      "step": 20763
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875996010164551e-06,
+      "loss": 0.2973,
+      "step": 20764
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875110068668051e-06,
+      "loss": 0.3167,
+      "step": 20765
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.874224204100511e-06,
+      "loss": 0.3067,
+      "step": 20766
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873338416473061e-06,
+      "loss": 0.3512,
+      "step": 20767
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.872452705796822e-06,
+      "loss": 0.3187,
+      "step": 20768
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.871567072082919e-06,
+      "loss": 0.3283,
+      "step": 20769
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87068151534248e-06,
+      "loss": 0.3232,
+      "step": 20770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.869796035586625e-06,
+      "loss": 0.3192,
+      "step": 20771
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868910632826471e-06,
+      "loss": 0.3662,
+      "step": 20772
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868025307073148e-06,
+      "loss": 0.2865,
+      "step": 20773
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.867140058337765e-06,
+      "loss": 0.328,
+      "step": 20774
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866254886631448e-06,
+      "loss": 0.3438,
+      "step": 20775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.865369791965316e-06,
+      "loss": 0.328,
+      "step": 20776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8644847743504786e-06,
+      "loss": 0.3487,
+      "step": 20777
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863599833798057e-06,
+      "loss": 0.3202,
+      "step": 20778
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862714970319166e-06,
+      "loss": 0.3464,
+      "step": 20779
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.861830183924921e-06,
+      "loss": 0.3126,
+      "step": 20780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860945474626433e-06,
+      "loss": 0.2961,
+      "step": 20781
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86006084243481e-06,
+      "loss": 0.3171,
+      "step": 20782
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85917628736117e-06,
+      "loss": 0.34,
+      "step": 20783
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858291809416617e-06,
+      "loss": 0.3289,
+      "step": 20784
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857407408612261e-06,
+      "loss": 0.2998,
+      "step": 20785
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.856523084959217e-06,
+      "loss": 0.3504,
+      "step": 20786
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855638838468584e-06,
+      "loss": 0.3254,
+      "step": 20787
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854754669151471e-06,
+      "loss": 0.33,
+      "step": 20788
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853870577018986e-06,
+      "loss": 0.3141,
+      "step": 20789
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8529865620822295e-06,
+      "loss": 0.3243,
+      "step": 20790
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852102624352307e-06,
+      "loss": 0.3099,
+      "step": 20791
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851218763840318e-06,
+      "loss": 0.3412,
+      "step": 20792
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8503349805573685e-06,
+      "loss": 0.3258,
+      "step": 20793
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.849451274514553e-06,
+      "loss": 0.3032,
+      "step": 20794
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848567645722977e-06,
+      "loss": 0.309,
+      "step": 20795
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.847684094193733e-06,
+      "loss": 0.3346,
+      "step": 20796
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84680061993792e-06,
+      "loss": 0.3177,
+      "step": 20797
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845917222966641e-06,
+      "loss": 0.317,
+      "step": 20798
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845033903290982e-06,
+      "loss": 0.3247,
+      "step": 20799
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8441506609220425e-06,
+      "loss": 0.3136,
+      "step": 20800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.843267495870918e-06,
+      "loss": 0.3239,
+      "step": 20801
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.842384408148696e-06,
+      "loss": 0.3245,
+      "step": 20802
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.841501397766474e-06,
+      "loss": 0.3197,
+      "step": 20803
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840618464735338e-06,
+      "loss": 0.316,
+      "step": 20804
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8397356090663775e-06,
+      "loss": 0.3154,
+      "step": 20805
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838852830770682e-06,
+      "loss": 0.3054,
+      "step": 20806
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.837970129859343e-06,
+      "loss": 0.3193,
+      "step": 20807
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.837087506343441e-06,
+      "loss": 0.3311,
+      "step": 20808
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.836204960234066e-06,
+      "loss": 0.3251,
+      "step": 20809
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835322491542303e-06,
+      "loss": 0.3014,
+      "step": 20810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.834440100279234e-06,
+      "loss": 0.2947,
+      "step": 20811
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833557786455942e-06,
+      "loss": 0.3213,
+      "step": 20812
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8326755500835126e-06,
+      "loss": 0.3309,
+      "step": 20813
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831793391173023e-06,
+      "loss": 0.3431,
+      "step": 20814
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830911309735552e-06,
+      "loss": 0.304,
+      "step": 20815
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8300293057821795e-06,
+      "loss": 0.3258,
+      "step": 20816
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.829147379323988e-06,
+      "loss": 0.3237,
+      "step": 20817
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828265530372047e-06,
+      "loss": 0.3217,
+      "step": 20818
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827383758937437e-06,
+      "loss": 0.3505,
+      "step": 20819
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.826502065031237e-06,
+      "loss": 0.3031,
+      "step": 20820
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825620448664513e-06,
+      "loss": 0.3095,
+      "step": 20821
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8247389098483446e-06,
+      "loss": 0.3132,
+      "step": 20822
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823857448593799e-06,
+      "loss": 0.3333,
+      "step": 20823
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822976064911953e-06,
+      "loss": 0.3066,
+      "step": 20824
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822094758813869e-06,
+      "loss": 0.3099,
+      "step": 20825
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821213530310625e-06,
+      "loss": 0.2995,
+      "step": 20826
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820332379413282e-06,
+      "loss": 0.318,
+      "step": 20827
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819451306132909e-06,
+      "loss": 0.2855,
+      "step": 20828
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81857031048058e-06,
+      "loss": 0.2634,
+      "step": 20829
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817689392467348e-06,
+      "loss": 0.2968,
+      "step": 20830
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816808552104285e-06,
+      "loss": 0.3159,
+      "step": 20831
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.815927789402456e-06,
+      "loss": 0.3502,
+      "step": 20832
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.815047104372916e-06,
+      "loss": 0.3523,
+      "step": 20833
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8141664970267355e-06,
+      "loss": 0.3182,
+      "step": 20834
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8132859673749688e-06,
+      "loss": 0.3444,
+      "step": 20835
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8124055154286744e-06,
+      "loss": 0.3112,
+      "step": 20836
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.811525141198913e-06,
+      "loss": 0.3216,
+      "step": 20837
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.810644844696746e-06,
+      "loss": 0.3236,
+      "step": 20838
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8097646259332235e-06,
+      "loss": 0.3099,
+      "step": 20839
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8088844849194027e-06,
+      "loss": 0.3192,
+      "step": 20840
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8080044216663435e-06,
+      "loss": 0.3226,
+      "step": 20841
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8071244361850924e-06,
+      "loss": 0.3278,
+      "step": 20842
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8062445284867054e-06,
+      "loss": 0.3012,
+      "step": 20843
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.805364698582237e-06,
+      "loss": 0.3327,
+      "step": 20844
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.804484946482735e-06,
+      "loss": 0.3186,
+      "step": 20845
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8036052721992454e-06,
+      "loss": 0.355,
+      "step": 20846
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8027256757428243e-06,
+      "loss": 0.3231,
+      "step": 20847
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8018461571245124e-06,
+      "loss": 0.3189,
+      "step": 20848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.80096671635536e-06,
+      "loss": 0.3322,
+      "step": 20849
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8000873534464164e-06,
+      "loss": 0.2948,
+      "step": 20850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7992080684087184e-06,
+      "loss": 0.3419,
+      "step": 20851
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7983288612533155e-06,
+      "loss": 0.3213,
+      "step": 20852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7974497319912496e-06,
+      "loss": 0.319,
+      "step": 20853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7965706806335643e-06,
+      "loss": 0.3223,
+      "step": 20854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7956917071913003e-06,
+      "loss": 0.3406,
+      "step": 20855
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7948128116754923e-06,
+      "loss": 0.3123,
+      "step": 20856
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7939339940971852e-06,
+      "loss": 0.305,
+      "step": 20857
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7930552544674125e-06,
+      "loss": 0.3237,
+      "step": 20858
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.792176592797213e-06,
+      "loss": 0.3138,
+      "step": 20859
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7912980090976272e-06,
+      "loss": 0.3024,
+      "step": 20860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7904195033796807e-06,
+      "loss": 0.3059,
+      "step": 20861
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.789541075654415e-06,
+      "loss": 0.284,
+      "step": 20862
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7886627259328645e-06,
+      "loss": 0.3184,
+      "step": 20863
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7877844542260533e-06,
+      "loss": 0.3056,
+      "step": 20864
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7869062605450225e-06,
+      "loss": 0.2846,
+      "step": 20865
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7860281449007953e-06,
+      "loss": 0.2759,
+      "step": 20866
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7851501073044005e-06,
+      "loss": 0.3542,
+      "step": 20867
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7842721477668674e-06,
+      "loss": 0.3133,
+      "step": 20868
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7833942662992286e-06,
+      "loss": 0.2998,
+      "step": 20869
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7825164629125022e-06,
+      "loss": 0.3028,
+      "step": 20870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7816387376177167e-06,
+      "loss": 0.3043,
+      "step": 20871
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.780761090425901e-06,
+      "loss": 0.3327,
+      "step": 20872
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779883521348071e-06,
+      "loss": 0.3162,
+      "step": 20873
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779006030395251e-06,
+      "loss": 0.3316,
+      "step": 20874
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.778128617578469e-06,
+      "loss": 0.3061,
+      "step": 20875
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777251282908739e-06,
+      "loss": 0.2758,
+      "step": 20876
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7763740263970784e-06,
+      "loss": 0.3388,
+      "step": 20877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7754968480545118e-06,
+      "loss": 0.3295,
+      "step": 20878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7746197478920498e-06,
+      "loss": 0.3342,
+      "step": 20879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.773742725920713e-06,
+      "loss": 0.3241,
+      "step": 20880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.772865782151519e-06,
+      "loss": 0.3023,
+      "step": 20881
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771988916595476e-06,
+      "loss": 0.3331,
+      "step": 20882
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771112129263602e-06,
+      "loss": 0.3365,
+      "step": 20883
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7702354201669113e-06,
+      "loss": 0.3156,
+      "step": 20884
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7693587893164085e-06,
+      "loss": 0.3247,
+      "step": 20885
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7684822367231132e-06,
+      "loss": 0.3201,
+      "step": 20886
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7676057623980255e-06,
+      "loss": 0.2974,
+      "step": 20887
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.766729366352162e-06,
+      "loss": 0.3248,
+      "step": 20888
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.765853048596524e-06,
+      "loss": 0.302,
+      "step": 20889
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7649768091421203e-06,
+      "loss": 0.2931,
+      "step": 20890
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.764100647999961e-06,
+      "loss": 0.2983,
+      "step": 20891
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.763224565181043e-06,
+      "loss": 0.3092,
+      "step": 20892
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.762348560696373e-06,
+      "loss": 0.3415,
+      "step": 20893
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7614726345569584e-06,
+      "loss": 0.3196,
+      "step": 20894
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7605967867737934e-06,
+      "loss": 0.3244,
+      "step": 20895
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7597210173578868e-06,
+      "loss": 0.3124,
+      "step": 20896
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7588453263202295e-06,
+      "loss": 0.3123,
+      "step": 20897
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7579697136718276e-06,
+      "loss": 0.2924,
+      "step": 20898
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7570941794236725e-06,
+      "loss": 0.2991,
+      "step": 20899
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7562187235867676e-06,
+      "loss": 0.2985,
+      "step": 20900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.755343346172101e-06,
+      "loss": 0.3254,
+      "step": 20901
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7544680471906727e-06,
+      "loss": 0.3014,
+      "step": 20902
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7535928266534795e-06,
+      "loss": 0.3131,
+      "step": 20903
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7527176845715053e-06,
+      "loss": 0.3203,
+      "step": 20904
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7518426209557478e-06,
+      "loss": 0.3172,
+      "step": 20905
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7509676358171998e-06,
+      "loss": 0.3318,
+      "step": 20906
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7500927291668453e-06,
+      "loss": 0.2976,
+      "step": 20907
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7492179010156795e-06,
+      "loss": 0.3189,
+      "step": 20908
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7483431513746873e-06,
+      "loss": 0.3452,
+      "step": 20909
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.747468480254851e-06,
+      "loss": 0.2844,
+      "step": 20910
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7465938876671615e-06,
+      "loss": 0.2835,
+      "step": 20911
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.745719373622606e-06,
+      "loss": 0.31,
+      "step": 20912
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7448449381321616e-06,
+      "loss": 0.2995,
+      "step": 20913
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.743970581206816e-06,
+      "loss": 0.2707,
+      "step": 20914
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.743096302857553e-06,
+      "loss": 0.3378,
+      "step": 20915
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7422221030953477e-06,
+      "loss": 0.3244,
+      "step": 20916
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741347981931184e-06,
+      "loss": 0.2592,
+      "step": 20917
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7404739393760424e-06,
+      "loss": 0.3207,
+      "step": 20918
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7395999754408995e-06,
+      "loss": 0.3222,
+      "step": 20919
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.738726090136727e-06,
+      "loss": 0.3453,
+      "step": 20920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.73785228347451e-06,
+      "loss": 0.3021,
+      "step": 20921
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.736978555465215e-06,
+      "loss": 0.3208,
+      "step": 20922
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7361049061198208e-06,
+      "loss": 0.3301,
+      "step": 20923
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7352313354493e-06,
+      "loss": 0.3333,
+      "step": 20924
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7343578434646266e-06,
+      "loss": 0.3128,
+      "step": 20925
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733484430176767e-06,
+      "loss": 0.3396,
+      "step": 20926
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7326110955966942e-06,
+      "loss": 0.3035,
+      "step": 20927
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.73173783973538e-06,
+      "loss": 0.3449,
+      "step": 20928
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7308646626037893e-06,
+      "loss": 0.3061,
+      "step": 20929
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729991564212887e-06,
+      "loss": 0.3521,
+      "step": 20930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7291185445736445e-06,
+      "loss": 0.3148,
+      "step": 20931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.728245603697022e-06,
+      "loss": 0.3421,
+      "step": 20932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.727372741593984e-06,
+      "loss": 0.3144,
+      "step": 20933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7264999582755003e-06,
+      "loss": 0.3477,
+      "step": 20934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7256272537525253e-06,
+      "loss": 0.3018,
+      "step": 20935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7247546280360227e-06,
+      "loss": 0.3479,
+      "step": 20936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.723882081136958e-06,
+      "loss": 0.3137,
+      "step": 20937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7230096130662805e-06,
+      "loss": 0.3322,
+      "step": 20938
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7221372238349597e-06,
+      "loss": 0.2846,
+      "step": 20939
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.721264913453945e-06,
+      "loss": 0.3371,
+      "step": 20940
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.720392681934192e-06,
+      "loss": 0.2784,
+      "step": 20941
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7195205292866588e-06,
+      "loss": 0.3175,
+      "step": 20942
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7186484555223025e-06,
+      "loss": 0.2838,
+      "step": 20943
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7177764606520696e-06,
+      "loss": 0.3222,
+      "step": 20944
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716904544686917e-06,
+      "loss": 0.3085,
+      "step": 20945
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716032707637798e-06,
+      "loss": 0.2954,
+      "step": 20946
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7151609495156573e-06,
+      "loss": 0.3057,
+      "step": 20947
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.714289270331447e-06,
+      "loss": 0.3085,
+      "step": 20948
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.713417670096119e-06,
+      "loss": 0.3582,
+      "step": 20949
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.712546148820616e-06,
+      "loss": 0.3093,
+      "step": 20950
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.711674706515883e-06,
+      "loss": 0.2508,
+      "step": 20951
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7108033431928716e-06,
+      "loss": 0.3232,
+      "step": 20952
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7099320588625187e-06,
+      "loss": 0.3178,
+      "step": 20953
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7090608535357708e-06,
+      "loss": 0.3193,
+      "step": 20954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7081897272235756e-06,
+      "loss": 0.3227,
+      "step": 20955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7073186799368656e-06,
+      "loss": 0.3321,
+      "step": 20956
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7064477116865847e-06,
+      "loss": 0.3173,
+      "step": 20957
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7055768224836775e-06,
+      "loss": 0.3002,
+      "step": 20958
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.704706012339074e-06,
+      "loss": 0.2981,
+      "step": 20959
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7038352812637203e-06,
+      "loss": 0.3435,
+      "step": 20960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7029646292685428e-06,
+      "loss": 0.3533,
+      "step": 20961
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702094056364487e-06,
+      "loss": 0.3193,
+      "step": 20962
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.701223562562478e-06,
+      "loss": 0.2884,
+      "step": 20963
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7003531478734546e-06,
+      "loss": 0.3251,
+      "step": 20964
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6994828123083516e-06,
+      "loss": 0.3085,
+      "step": 20965
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.698612555878095e-06,
+      "loss": 0.326,
+      "step": 20966
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6977423785936163e-06,
+      "loss": 0.3451,
+      "step": 20967
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69687228046585e-06,
+      "loss": 0.2906,
+      "step": 20968
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.696002261505718e-06,
+      "loss": 0.3512,
+      "step": 20969
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.695132321724153e-06,
+      "loss": 0.3258,
+      "step": 20970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69426246113208e-06,
+      "loss": 0.3107,
+      "step": 20971
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6933926797404198e-06,
+      "loss": 0.3695,
+      "step": 20972
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6925229775601014e-06,
+      "loss": 0.3023,
+      "step": 20973
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6916533546020504e-06,
+      "loss": 0.373,
+      "step": 20974
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6907838108771833e-06,
+      "loss": 0.3098,
+      "step": 20975
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6899143463964247e-06,
+      "loss": 0.3056,
+      "step": 20976
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6890449611706993e-06,
+      "loss": 0.299,
+      "step": 20977
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6881756552109192e-06,
+      "loss": 0.3159,
+      "step": 20978
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.687306428528006e-06,
+      "loss": 0.3116,
+      "step": 20979
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.686437281132881e-06,
+      "loss": 0.3101,
+      "step": 20980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6855682130364577e-06,
+      "loss": 0.3165,
+      "step": 20981
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.684699224249647e-06,
+      "loss": 0.3031,
+      "step": 20982
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.683830314783372e-06,
+      "loss": 0.2982,
+      "step": 20983
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6829614846485386e-06,
+      "loss": 0.3144,
+      "step": 20984
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.682092733856063e-06,
+      "loss": 0.2877,
+      "step": 20985
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6812240624168595e-06,
+      "loss": 0.3096,
+      "step": 20986
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.680355470341832e-06,
+      "loss": 0.33,
+      "step": 20987
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6794869576418935e-06,
+      "loss": 0.2876,
+      "step": 20988
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6786185243279573e-06,
+      "loss": 0.3122,
+      "step": 20989
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.677750170410922e-06,
+      "loss": 0.3287,
+      "step": 20990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6768818959017025e-06,
+      "loss": 0.3414,
+      "step": 20991
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6760137008111973e-06,
+      "loss": 0.2913,
+      "step": 20992
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.675145585150317e-06,
+      "loss": 0.3249,
+      "step": 20993
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.67427754892996e-06,
+      "loss": 0.3241,
+      "step": 20994
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6734095921610336e-06,
+      "loss": 0.289,
+      "step": 20995
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.672541714854434e-06,
+      "loss": 0.3423,
+      "step": 20996
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6716739170210645e-06,
+      "loss": 0.3583,
+      "step": 20997
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6708061986718258e-06,
+      "loss": 0.3585,
+      "step": 20998
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6699385598176185e-06,
+      "loss": 0.3252,
+      "step": 20999
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6690710004693354e-06,
+      "loss": 0.3448,
+      "step": 21000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6682035206378763e-06,
+      "loss": 0.3334,
+      "step": 21001
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.667336120334134e-06,
+      "loss": 0.2986,
+      "step": 21002
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6664687995690075e-06,
+      "loss": 0.3186,
+      "step": 21003
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6656015583533835e-06,
+      "loss": 0.3188,
+      "step": 21004
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.664734396698163e-06,
+      "loss": 0.3468,
+      "step": 21005
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.663867314614229e-06,
+      "loss": 0.2817,
+      "step": 21006
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.663000312112478e-06,
+      "loss": 0.3398,
+      "step": 21007
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6621333892038003e-06,
+      "loss": 0.3253,
+      "step": 21008
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.661266545899078e-06,
+      "loss": 0.3373,
+      "step": 21009
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.660399782209204e-06,
+      "loss": 0.3131,
+      "step": 21010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6595330981450674e-06,
+      "loss": 0.3368,
+      "step": 21011
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6586664937175463e-06,
+      "loss": 0.3121,
+      "step": 21012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.657799968937532e-06,
+      "loss": 0.3264,
+      "step": 21013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6569335238159064e-06,
+      "loss": 0.2992,
+      "step": 21014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6560671583635467e-06,
+      "loss": 0.3341,
+      "step": 21015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.655200872591339e-06,
+      "loss": 0.3067,
+      "step": 21016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6543346665101676e-06,
+      "loss": 0.3373,
+      "step": 21017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6534685401309043e-06,
+      "loss": 0.3477,
+      "step": 21018
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.652602493464432e-06,
+      "loss": 0.2976,
+      "step": 21019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.651736526521631e-06,
+      "loss": 0.3273,
+      "step": 21020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.650870639313371e-06,
+      "loss": 0.3042,
+      "step": 21021
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.650004831850532e-06,
+      "loss": 0.3112,
+      "step": 21022
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6491391041439917e-06,
+      "loss": 0.3089,
+      "step": 21023
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6482734562046183e-06,
+      "loss": 0.3234,
+      "step": 21024
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.647407888043284e-06,
+      "loss": 0.315,
+      "step": 21025
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.646542399670866e-06,
+      "loss": 0.3219,
+      "step": 21026
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.645676991098227e-06,
+      "loss": 0.3206,
+      "step": 21027
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6448116623362416e-06,
+      "loss": 0.3291,
+      "step": 21028
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6439464133957805e-06,
+      "loss": 0.3129,
+      "step": 21029
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643081244287705e-06,
+      "loss": 0.3083,
+      "step": 21030
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.642216155022885e-06,
+      "loss": 0.3253,
+      "step": 21031
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6413511456121906e-06,
+      "loss": 0.3403,
+      "step": 21032
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6404862160664778e-06,
+      "loss": 0.3307,
+      "step": 21033
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6396213663966173e-06,
+      "loss": 0.3004,
+      "step": 21034
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6387565966134665e-06,
+      "loss": 0.329,
+      "step": 21035
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6378919067278906e-06,
+      "loss": 0.3408,
+      "step": 21036
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637027296750747e-06,
+      "loss": 0.3216,
+      "step": 21037
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6361627666928965e-06,
+      "loss": 0.3137,
+      "step": 21038
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6352983165652013e-06,
+      "loss": 0.2831,
+      "step": 21039
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6344339463785115e-06,
+      "loss": 0.2991,
+      "step": 21040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6335696561436897e-06,
+      "loss": 0.3054,
+      "step": 21041
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6327054458715915e-06,
+      "loss": 0.3124,
+      "step": 21042
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.631841315573066e-06,
+      "loss": 0.3012,
+      "step": 21043
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630977265258975e-06,
+      "loss": 0.2998,
+      "step": 21044
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630113294940165e-06,
+      "loss": 0.3228,
+      "step": 21045
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.629249404627485e-06,
+      "loss": 0.2964,
+      "step": 21046
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6283855943317903e-06,
+      "loss": 0.3318,
+      "step": 21047
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6275218640639333e-06,
+      "loss": 0.3057,
+      "step": 21048
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.626658213834754e-06,
+      "loss": 0.3389,
+      "step": 21049
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.625794643655105e-06,
+      "loss": 0.3023,
+      "step": 21050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6249311535358356e-06,
+      "loss": 0.2735,
+      "step": 21051
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.624067743487785e-06,
+      "loss": 0.3255,
+      "step": 21052
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.623204413521799e-06,
+      "loss": 0.278,
+      "step": 21053
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6223411636487272e-06,
+      "loss": 0.2993,
+      "step": 21054
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6214779938794075e-06,
+      "loss": 0.284,
+      "step": 21055
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6206149042246784e-06,
+      "loss": 0.3504,
+      "step": 21056
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6197518946953868e-06,
+      "loss": 0.3618,
+      "step": 21057
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6188889653023652e-06,
+      "loss": 0.3258,
+      "step": 21058
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.618026116056456e-06,
+      "loss": 0.3328,
+      "step": 21059
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6171633469684984e-06,
+      "loss": 0.2786,
+      "step": 21060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6163006580493245e-06,
+      "loss": 0.329,
+      "step": 21061
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6154380493097705e-06,
+      "loss": 0.2929,
+      "step": 21062
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.614575520760677e-06,
+      "loss": 0.3321,
+      "step": 21063
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.613713072412868e-06,
+      "loss": 0.3175,
+      "step": 21064
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6128507042771844e-06,
+      "loss": 0.2828,
+      "step": 21065
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.61198841636445e-06,
+      "loss": 0.2911,
+      "step": 21066
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6111262086855027e-06,
+      "loss": 0.3242,
+      "step": 21067
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6102640812511648e-06,
+      "loss": 0.2983,
+      "step": 21068
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6094020340722715e-06,
+      "loss": 0.3336,
+      "step": 21069
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6085400671596416e-06,
+      "loss": 0.3397,
+      "step": 21070
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607678180524108e-06,
+      "loss": 0.315,
+      "step": 21071
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606816374176494e-06,
+      "loss": 0.326,
+      "step": 21072
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6059546481276276e-06,
+      "loss": 0.3049,
+      "step": 21073
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605093002388326e-06,
+      "loss": 0.3228,
+      "step": 21074
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6042314369694176e-06,
+      "loss": 0.3091,
+      "step": 21075
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.603369951881717e-06,
+      "loss": 0.3295,
+      "step": 21076
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.602508547136051e-06,
+      "loss": 0.3053,
+      "step": 21077
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6016472227432332e-06,
+      "loss": 0.2984,
+      "step": 21078
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6007859787140874e-06,
+      "loss": 0.3098,
+      "step": 21079
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5999248150594257e-06,
+      "loss": 0.3064,
+      "step": 21080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.599063731790067e-06,
+      "loss": 0.309,
+      "step": 21081
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5982027289168287e-06,
+      "loss": 0.3118,
+      "step": 21082
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5973418064505196e-06,
+      "loss": 0.2799,
+      "step": 21083
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.596480964401955e-06,
+      "loss": 0.3422,
+      "step": 21084
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.595620202781953e-06,
+      "loss": 0.2903,
+      "step": 21085
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59475952160132e-06,
+      "loss": 0.3364,
+      "step": 21086
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.593898920870862e-06,
+      "loss": 0.3405,
+      "step": 21087
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.593038400601395e-06,
+      "loss": 0.3313,
+      "step": 21088
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5921779608037222e-06,
+      "loss": 0.3223,
+      "step": 21089
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.591317601488653e-06,
+      "loss": 0.338,
+      "step": 21090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.590457322666997e-06,
+      "loss": 0.3499,
+      "step": 21091
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.589597124349553e-06,
+      "loss": 0.341,
+      "step": 21092
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.588737006547127e-06,
+      "loss": 0.3433,
+      "step": 21093
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5878769692705273e-06,
+      "loss": 0.3561,
+      "step": 21094
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5870170125305493e-06,
+      "loss": 0.2979,
+      "step": 21095
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5861571363379988e-06,
+      "loss": 0.3026,
+      "step": 21096
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5852973407036717e-06,
+      "loss": 0.3044,
+      "step": 21097
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.584437625638372e-06,
+      "loss": 0.2812,
+      "step": 21098
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.583577991152892e-06,
+      "loss": 0.3068,
+      "step": 21099
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.582718437258035e-06,
+      "loss": 0.3259,
+      "step": 21100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.581858963964591e-06,
+      "loss": 0.3387,
+      "step": 21101
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5809995712833566e-06,
+      "loss": 0.3408,
+      "step": 21102
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.580140259225131e-06,
+      "loss": 0.2976,
+      "step": 21103
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5792810278007007e-06,
+      "loss": 0.3049,
+      "step": 21104
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.578421877020859e-06,
+      "loss": 0.3249,
+      "step": 21105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.577562806896403e-06,
+      "loss": 0.3029,
+      "step": 21106
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.576703817438113e-06,
+      "loss": 0.3367,
+      "step": 21107
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5758449086567872e-06,
+      "loss": 0.3206,
+      "step": 21108
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5749860805632053e-06,
+      "loss": 0.343,
+      "step": 21109
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5741273331681616e-06,
+      "loss": 0.3022,
+      "step": 21110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5732686664824354e-06,
+      "loss": 0.3532,
+      "step": 21111
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.572410080516814e-06,
+      "loss": 0.294,
+      "step": 21112
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.571551575282086e-06,
+      "loss": 0.3675,
+      "step": 21113
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5706931507890263e-06,
+      "loss": 0.3552,
+      "step": 21114
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5698348070484213e-06,
+      "loss": 0.3397,
+      "step": 21115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5689765440710543e-06,
+      "loss": 0.3654,
+      "step": 21116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5681183618676986e-06,
+      "loss": 0.2909,
+      "step": 21117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.56726026044914e-06,
+      "loss": 0.3016,
+      "step": 21118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566402239826153e-06,
+      "loss": 0.3229,
+      "step": 21119
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5655443000095115e-06,
+      "loss": 0.2633,
+      "step": 21120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5646864410099922e-06,
+      "loss": 0.3287,
+      "step": 21121
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5638286628383768e-06,
+      "loss": 0.3523,
+      "step": 21122
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.562970965505429e-06,
+      "loss": 0.3022,
+      "step": 21123
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5621133490219273e-06,
+      "loss": 0.3229,
+      "step": 21124
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5612558133986453e-06,
+      "loss": 0.327,
+      "step": 21125
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5603983586463476e-06,
+      "loss": 0.3591,
+      "step": 21126
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5595409847758076e-06,
+      "loss": 0.2847,
+      "step": 21127
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5586836917977963e-06,
+      "loss": 0.3182,
+      "step": 21128
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5578264797230786e-06,
+      "loss": 0.2988,
+      "step": 21129
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.556969348562419e-06,
+      "loss": 0.2901,
+      "step": 21130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.556112298326587e-06,
+      "loss": 0.3243,
+      "step": 21131
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5552553290263424e-06,
+      "loss": 0.3367,
+      "step": 21132
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.554398440672453e-06,
+      "loss": 0.3288,
+      "step": 21133
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.553541633275682e-06,
+      "loss": 0.3097,
+      "step": 21134
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5526849068467873e-06,
+      "loss": 0.2808,
+      "step": 21135
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5518282613965304e-06,
+      "loss": 0.3029,
+      "step": 21136
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5509716969356743e-06,
+      "loss": 0.3014,
+      "step": 21137
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5501152134749726e-06,
+      "loss": 0.3252,
+      "step": 21138
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.549258811025188e-06,
+      "loss": 0.3309,
+      "step": 21139
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5484024895970746e-06,
+      "loss": 0.3307,
+      "step": 21140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5475462492013836e-06,
+      "loss": 0.2963,
+      "step": 21141
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.546690089848872e-06,
+      "loss": 0.3424,
+      "step": 21142
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5458340115502955e-06,
+      "loss": 0.291,
+      "step": 21143
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.544978014316408e-06,
+      "loss": 0.3433,
+      "step": 21144
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.544122098157956e-06,
+      "loss": 0.3363,
+      "step": 21145
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5432662630856907e-06,
+      "loss": 0.2909,
+      "step": 21146
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5424105091103668e-06,
+      "loss": 0.3586,
+      "step": 21147
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5415548362427242e-06,
+      "loss": 0.3109,
+      "step": 21148
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5406992444935196e-06,
+      "loss": 0.3,
+      "step": 21149
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5398437338734935e-06,
+      "loss": 0.3211,
+      "step": 21150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538988304393388e-06,
+      "loss": 0.3401,
+      "step": 21151
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538132956063952e-06,
+      "loss": 0.3158,
+      "step": 21152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5372776888959304e-06,
+      "loss": 0.3271,
+      "step": 21153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.536422502900061e-06,
+      "loss": 0.2887,
+      "step": 21154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5355673980870853e-06,
+      "loss": 0.3269,
+      "step": 21155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5347123744677502e-06,
+      "loss": 0.3055,
+      "step": 21156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5338574320527853e-06,
+      "loss": 0.2986,
+      "step": 21157
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5330025708529335e-06,
+      "loss": 0.333,
+      "step": 21158
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5321477908789347e-06,
+      "loss": 0.2654,
+      "step": 21159
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.531293092141522e-06,
+      "loss": 0.3126,
+      "step": 21160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5304384746514273e-06,
+      "loss": 0.2975,
+      "step": 21161
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.529583938419391e-06,
+      "loss": 0.3152,
+      "step": 21162
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.528729483456139e-06,
+      "loss": 0.3306,
+      "step": 21163
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527875109772407e-06,
+      "loss": 0.3214,
+      "step": 21164
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527020817378931e-06,
+      "loss": 0.2967,
+      "step": 21165
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.526166606286432e-06,
+      "loss": 0.2769,
+      "step": 21166
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5253124765056426e-06,
+      "loss": 0.3035,
+      "step": 21167
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524458428047294e-06,
+      "loss": 0.2883,
+      "step": 21168
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5236044609221077e-06,
+      "loss": 0.2953,
+      "step": 21169
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.522750575140814e-06,
+      "loss": 0.3288,
+      "step": 21170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.521896770714134e-06,
+      "loss": 0.3166,
+      "step": 21171
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5210430476527967e-06,
+      "loss": 0.3577,
+      "step": 21172
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.520189405967518e-06,
+      "loss": 0.2806,
+      "step": 21173
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5193358456690262e-06,
+      "loss": 0.3321,
+      "step": 21174
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5184823667680357e-06,
+      "loss": 0.2912,
+      "step": 21175
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517628969275271e-06,
+      "loss": 0.3169,
+      "step": 21176
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5167756532014507e-06,
+      "loss": 0.2978,
+      "step": 21177
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5159224185572895e-06,
+      "loss": 0.2922,
+      "step": 21178
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5150692653535056e-06,
+      "loss": 0.3332,
+      "step": 21179
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5142161936008178e-06,
+      "loss": 0.3396,
+      "step": 21180
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.513363203309935e-06,
+      "loss": 0.2976,
+      "step": 21181
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.512510294491577e-06,
+      "loss": 0.3067,
+      "step": 21182
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5116574671564497e-06,
+      "loss": 0.3273,
+      "step": 21183
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.510804721315272e-06,
+      "loss": 0.3513,
+      "step": 21184
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509952056978746e-06,
+      "loss": 0.3236,
+      "step": 21185
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509099474157586e-06,
+      "loss": 0.3182,
+      "step": 21186
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5082469728625034e-06,
+      "loss": 0.3146,
+      "step": 21187
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5073945531041987e-06,
+      "loss": 0.2968,
+      "step": 21188
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5065422148933824e-06,
+      "loss": 0.3066,
+      "step": 21189
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.505689958240763e-06,
+      "loss": 0.3001,
+      "step": 21190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504837783157038e-06,
+      "loss": 0.3159,
+      "step": 21191
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.503985689652918e-06,
+      "loss": 0.3005,
+      "step": 21192
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.503133677739101e-06,
+      "loss": 0.317,
+      "step": 21193
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.502281747426286e-06,
+      "loss": 0.3046,
+      "step": 21194
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.501429898725175e-06,
+      "loss": 0.3181,
+      "step": 21195
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5005781316464725e-06,
+      "loss": 0.328,
+      "step": 21196
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4997264462008693e-06,
+      "loss": 0.3431,
+      "step": 21197
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498874842399067e-06,
+      "loss": 0.2726,
+      "step": 21198
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4980233202517633e-06,
+      "loss": 0.3154,
+      "step": 21199
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4971718797696475e-06,
+      "loss": 0.3097,
+      "step": 21200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4963205209634167e-06,
+      "loss": 0.3062,
+      "step": 21201
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4954692438437686e-06,
+      "loss": 0.3067,
+      "step": 21202
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4946180484213898e-06,
+      "loss": 0.3107,
+      "step": 21203
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49376693470697e-06,
+      "loss": 0.3235,
+      "step": 21204
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4929159027112056e-06,
+      "loss": 0.3536,
+      "step": 21205
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.492064952444779e-06,
+      "loss": 0.3077,
+      "step": 21206
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4912140839183807e-06,
+      "loss": 0.2846,
+      "step": 21207
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4903632971427003e-06,
+      "loss": 0.3218,
+      "step": 21208
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.489512592128419e-06,
+      "loss": 0.3429,
+      "step": 21209
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4886619688862246e-06,
+      "loss": 0.2898,
+      "step": 21210
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4878114274268025e-06,
+      "loss": 0.3207,
+      "step": 21211
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.486960967760832e-06,
+      "loss": 0.3316,
+      "step": 21212
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.486110589898999e-06,
+      "loss": 0.2466,
+      "step": 21213
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4852602938519806e-06,
+      "loss": 0.293,
+      "step": 21214
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.484410079630456e-06,
+      "loss": 0.3278,
+      "step": 21215
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4835599472451056e-06,
+      "loss": 0.3337,
+      "step": 21216
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4827098967066063e-06,
+      "loss": 0.296,
+      "step": 21217
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4818599280256405e-06,
+      "loss": 0.3292,
+      "step": 21218
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4810100412128743e-06,
+      "loss": 0.31,
+      "step": 21219
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4801602362789876e-06,
+      "loss": 0.3166,
+      "step": 21220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479310513234657e-06,
+      "loss": 0.3063,
+      "step": 21221
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478460872090549e-06,
+      "loss": 0.3263,
+      "step": 21222
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4776113128573396e-06,
+      "loss": 0.3295,
+      "step": 21223
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4767618355456976e-06,
+      "loss": 0.3112,
+      "step": 21224
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.47591244016629e-06,
+      "loss": 0.3243,
+      "step": 21225
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4750631267297865e-06,
+      "loss": 0.3357,
+      "step": 21226
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4742138952468595e-06,
+      "loss": 0.3235,
+      "step": 21227
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4733647457281684e-06,
+      "loss": 0.3304,
+      "step": 21228
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4725156781843806e-06,
+      "loss": 0.3372,
+      "step": 21229
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4716666926261653e-06,
+      "loss": 0.2984,
+      "step": 21230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.470817789064178e-06,
+      "loss": 0.3008,
+      "step": 21231
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4699689675090863e-06,
+      "loss": 0.2877,
+      "step": 21232
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.469120227971553e-06,
+      "loss": 0.3093,
+      "step": 21233
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.468271570462235e-06,
+      "loss": 0.3422,
+      "step": 21234
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.467422994991788e-06,
+      "loss": 0.2914,
+      "step": 21235
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4665745015708783e-06,
+      "loss": 0.3228,
+      "step": 21236
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4657260902101553e-06,
+      "loss": 0.3383,
+      "step": 21237
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4648777609202787e-06,
+      "loss": 0.2974,
+      "step": 21238
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4640295137119062e-06,
+      "loss": 0.3131,
+      "step": 21239
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4631813485956868e-06,
+      "loss": 0.331,
+      "step": 21240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.462333265582275e-06,
+      "loss": 0.3021,
+      "step": 21241
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.461485264682326e-06,
+      "loss": 0.3185,
+      "step": 21242
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.460637345906487e-06,
+      "loss": 0.32,
+      "step": 21243
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.459789509265411e-06,
+      "loss": 0.3511,
+      "step": 21244
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4589417547697447e-06,
+      "loss": 0.2957,
+      "step": 21245
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4580940824301347e-06,
+      "loss": 0.3077,
+      "step": 21246
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.457246492257228e-06,
+      "loss": 0.2869,
+      "step": 21247
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4563989842616753e-06,
+      "loss": 0.3271,
+      "step": 21248
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4555515584541145e-06,
+      "loss": 0.318,
+      "step": 21249
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4547042148451927e-06,
+      "loss": 0.3163,
+      "step": 21250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453856953445557e-06,
+      "loss": 0.3052,
+      "step": 21251
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.45300977426584e-06,
+      "loss": 0.3044,
+      "step": 21252
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.452162677316687e-06,
+      "loss": 0.3162,
+      "step": 21253
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4513156626087407e-06,
+      "loss": 0.2939,
+      "step": 21254
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4504687301526376e-06,
+      "loss": 0.3387,
+      "step": 21255
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4496218799590097e-06,
+      "loss": 0.3119,
+      "step": 21256
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.448775112038497e-06,
+      "loss": 0.3006,
+      "step": 21257
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447928426401741e-06,
+      "loss": 0.2924,
+      "step": 21258
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4470818230593662e-06,
+      "loss": 0.2701,
+      "step": 21259
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.446235302022012e-06,
+      "loss": 0.2931,
+      "step": 21260
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4453888633003117e-06,
+      "loss": 0.3229,
+      "step": 21261
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.444542506904892e-06,
+      "loss": 0.3252,
+      "step": 21262
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.443696232846384e-06,
+      "loss": 0.3111,
+      "step": 21263
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4428500411354237e-06,
+      "loss": 0.3453,
+      "step": 21264
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4420039317826324e-06,
+      "loss": 0.3245,
+      "step": 21265
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4411579047986364e-06,
+      "loss": 0.3104,
+      "step": 21266
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.440311960194068e-06,
+      "loss": 0.3218,
+      "step": 21267
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4394660979795457e-06,
+      "loss": 0.3289,
+      "step": 21268
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4386203181656964e-06,
+      "loss": 0.2814,
+      "step": 21269
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.437774620763147e-06,
+      "loss": 0.3165,
+      "step": 21270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.436929005782512e-06,
+      "loss": 0.2908,
+      "step": 21271
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.436083473234416e-06,
+      "loss": 0.3457,
+      "step": 21272
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4352380231294835e-06,
+      "loss": 0.3008,
+      "step": 21273
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434392655478325e-06,
+      "loss": 0.2977,
+      "step": 21274
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4335473702915656e-06,
+      "loss": 0.3263,
+      "step": 21275
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4327021675798166e-06,
+      "loss": 0.2878,
+      "step": 21276
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4318570473536984e-06,
+      "loss": 0.3063,
+      "step": 21277
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.431012009623821e-06,
+      "loss": 0.2874,
+      "step": 21278
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4301670544008036e-06,
+      "loss": 0.3073,
+      "step": 21279
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4293221816952525e-06,
+      "loss": 0.3087,
+      "step": 21280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4284773915177837e-06,
+      "loss": 0.3206,
+      "step": 21281
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.42763268387901e-06,
+      "loss": 0.2959,
+      "step": 21282
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4267880587895343e-06,
+      "loss": 0.3059,
+      "step": 21283
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4259435162599684e-06,
+      "loss": 0.3488,
+      "step": 21284
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4250990563009233e-06,
+      "loss": 0.3037,
+      "step": 21285
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.424254678922999e-06,
+      "loss": 0.3131,
+      "step": 21286
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423410384136808e-06,
+      "loss": 0.3276,
+      "step": 21287
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4225661719529513e-06,
+      "loss": 0.3015,
+      "step": 21288
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4217220423820274e-06,
+      "loss": 0.2983,
+      "step": 21289
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.420877995434644e-06,
+      "loss": 0.3265,
+      "step": 21290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4200340311214007e-06,
+      "loss": 0.3435,
+      "step": 21291
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419190149452902e-06,
+      "loss": 0.3093,
+      "step": 21292
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.418346350439742e-06,
+      "loss": 0.2956,
+      "step": 21293
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4175026340925188e-06,
+      "loss": 0.3118,
+      "step": 21294
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4166590004218336e-06,
+      "loss": 0.3026,
+      "step": 21295
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4158154494382777e-06,
+      "loss": 0.289,
+      "step": 21296
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414971981152453e-06,
+      "loss": 0.2947,
+      "step": 21297
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414128595574947e-06,
+      "loss": 0.313,
+      "step": 21298
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4132852927163527e-06,
+      "loss": 0.2944,
+      "step": 21299
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4124420725872643e-06,
+      "loss": 0.3083,
+      "step": 21300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.411598935198275e-06,
+      "loss": 0.301,
+      "step": 21301
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4107558805599685e-06,
+      "loss": 0.2848,
+      "step": 21302
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4099129086829377e-06,
+      "loss": 0.3209,
+      "step": 21303
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4090700195777725e-06,
+      "loss": 0.3114,
+      "step": 21304
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.408227213255054e-06,
+      "loss": 0.3299,
+      "step": 21305
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4073844897253716e-06,
+      "loss": 0.2782,
+      "step": 21306
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4065418489993118e-06,
+      "loss": 0.3429,
+      "step": 21307
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4056992910874552e-06,
+      "loss": 0.2975,
+      "step": 21308
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4048568160003827e-06,
+      "loss": 0.3332,
+      "step": 21309
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4040144237486804e-06,
+      "loss": 0.3199,
+      "step": 21310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.403172114342924e-06,
+      "loss": 0.3352,
+      "step": 21311
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4023298877936937e-06,
+      "loss": 0.3329,
+      "step": 21312
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4014877441115745e-06,
+      "loss": 0.3117,
+      "step": 21313
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4006456833071345e-06,
+      "loss": 0.2791,
+      "step": 21314
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.399803705390955e-06,
+      "loss": 0.328,
+      "step": 21315
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3989618103736133e-06,
+      "loss": 0.3111,
+      "step": 21316
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3981199982656774e-06,
+      "loss": 0.3095,
+      "step": 21317
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3972782690777283e-06,
+      "loss": 0.3123,
+      "step": 21318
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.396436622820334e-06,
+      "loss": 0.3213,
+      "step": 21319
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395595059504061e-06,
+      "loss": 0.3306,
+      "step": 21320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3947535791394847e-06,
+      "loss": 0.3158,
+      "step": 21321
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.393912181737177e-06,
+      "loss": 0.3239,
+      "step": 21322
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3930708673076983e-06,
+      "loss": 0.3381,
+      "step": 21323
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.39222963586162e-06,
+      "loss": 0.3081,
+      "step": 21324
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3913884874095105e-06,
+      "loss": 0.2782,
+      "step": 21325
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3905474219619295e-06,
+      "loss": 0.3247,
+      "step": 21326
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3897064395294423e-06,
+      "loss": 0.3397,
+      "step": 21327
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3888655401226146e-06,
+      "loss": 0.33,
+      "step": 21328
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.388024723752007e-06,
+      "loss": 0.3342,
+      "step": 21329
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3871839904281765e-06,
+      "loss": 0.3008,
+      "step": 21330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.386343340161684e-06,
+      "loss": 0.2995,
+      "step": 21331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.385502772963094e-06,
+      "loss": 0.2994,
+      "step": 21332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3846622888429557e-06,
+      "loss": 0.3441,
+      "step": 21333
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3838218878118302e-06,
+      "loss": 0.302,
+      "step": 21334
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382981569880276e-06,
+      "loss": 0.3053,
+      "step": 21335
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3821413350588395e-06,
+      "loss": 0.3082,
+      "step": 21336
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.381301183358079e-06,
+      "loss": 0.3107,
+      "step": 21337
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.380461114788549e-06,
+      "loss": 0.2985,
+      "step": 21338
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3796211293607984e-06,
+      "loss": 0.2873,
+      "step": 21339
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.378781227085375e-06,
+      "loss": 0.2895,
+      "step": 21340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3779414079728324e-06,
+      "loss": 0.3129,
+      "step": 21341
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3771016720337136e-06,
+      "loss": 0.3204,
+      "step": 21342
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.376262019278569e-06,
+      "loss": 0.3557,
+      "step": 21343
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3754224497179467e-06,
+      "loss": 0.3263,
+      "step": 21344
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3745829633623874e-06,
+      "loss": 0.3357,
+      "step": 21345
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3737435602224355e-06,
+      "loss": 0.2517,
+      "step": 21346
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.37290424030864e-06,
+      "loss": 0.334,
+      "step": 21347
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.372065003631535e-06,
+      "loss": 0.2943,
+      "step": 21348
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3712258502016684e-06,
+      "loss": 0.3159,
+      "step": 21349
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3703867800295766e-06,
+      "loss": 0.3246,
+      "step": 21350
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3695477931257934e-06,
+      "loss": 0.3093,
+      "step": 21351
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3687088895008624e-06,
+      "loss": 0.3117,
+      "step": 21352
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.367870069165322e-06,
+      "loss": 0.3396,
+      "step": 21353
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.367031332129702e-06,
+      "loss": 0.2969,
+      "step": 21354
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3661926784045408e-06,
+      "loss": 0.3319,
+      "step": 21355
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3653541080003727e-06,
+      "loss": 0.3034,
+      "step": 21356
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3645156209277264e-06,
+      "loss": 0.3489,
+      "step": 21357
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3636772171971354e-06,
+      "loss": 0.325,
+      "step": 21358
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.362838896819133e-06,
+      "loss": 0.3126,
+      "step": 21359
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3620006598042465e-06,
+      "loss": 0.3045,
+      "step": 21360
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3611625061629995e-06,
+      "loss": 0.3342,
+      "step": 21361
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.360324435905923e-06,
+      "loss": 0.3114,
+      "step": 21362
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3594864490435473e-06,
+      "loss": 0.3376,
+      "step": 21363
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3586485455863905e-06,
+      "loss": 0.3303,
+      "step": 21364
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.35781072554498e-06,
+      "loss": 0.2919,
+      "step": 21365
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.356972988929842e-06,
+      "loss": 0.3296,
+      "step": 21366
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3561353357514913e-06,
+      "loss": 0.2649,
+      "step": 21367
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3552977660204533e-06,
+      "loss": 0.3015,
+      "step": 21368
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3544602797472514e-06,
+      "loss": 0.3382,
+      "step": 21369
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3536228769423995e-06,
+      "loss": 0.3354,
+      "step": 21370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3527855576164146e-06,
+      "loss": 0.3381,
+      "step": 21371
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.351948321779818e-06,
+      "loss": 0.3212,
+      "step": 21372
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3511111694431187e-06,
+      "loss": 0.3049,
+      "step": 21373
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.350274100616837e-06,
+      "loss": 0.3265,
+      "step": 21374
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3494371153114868e-06,
+      "loss": 0.3387,
+      "step": 21375
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3486002135375773e-06,
+      "loss": 0.3208,
+      "step": 21376
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.34776339530562e-06,
+      "loss": 0.3129,
+      "step": 21377
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.346926660626131e-06,
+      "loss": 0.3099,
+      "step": 21378
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.346090009509613e-06,
+      "loss": 0.356,
+      "step": 21379
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3452534419665793e-06,
+      "loss": 0.3417,
+      "step": 21380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3444169580075327e-06,
+      "loss": 0.3243,
+      "step": 21381
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3435805576429846e-06,
+      "loss": 0.3355,
+      "step": 21382
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3427442408834344e-06,
+      "loss": 0.31,
+      "step": 21383
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.341908007739393e-06,
+      "loss": 0.2913,
+      "step": 21384
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3410718582213562e-06,
+      "loss": 0.3082,
+      "step": 21385
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3402357923398297e-06,
+      "loss": 0.2886,
+      "step": 21386
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.339399810105318e-06,
+      "loss": 0.315,
+      "step": 21387
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3385639115283154e-06,
+      "loss": 0.3304,
+      "step": 21388
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3377280966193214e-06,
+      "loss": 0.322,
+      "step": 21389
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3368923653888406e-06,
+      "loss": 0.2938,
+      "step": 21390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.33605671784736e-06,
+      "loss": 0.307,
+      "step": 21391
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3352211540053847e-06,
+      "loss": 0.3266,
+      "step": 21392
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3343856738734047e-06,
+      "loss": 0.2761,
+      "step": 21393
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.333550277461911e-06,
+      "loss": 0.2814,
+      "step": 21394
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3327149647813984e-06,
+      "loss": 0.3296,
+      "step": 21395
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3318797358423638e-06,
+      "loss": 0.3589,
+      "step": 21396
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.331044590655289e-06,
+      "loss": 0.3337,
+      "step": 21397
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.330209529230667e-06,
+      "loss": 0.3378,
+      "step": 21398
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3293745515789876e-06,
+      "loss": 0.2897,
+      "step": 21399
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.328539657710741e-06,
+      "loss": 0.3372,
+      "step": 21400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.327704847636406e-06,
+      "loss": 0.3407,
+      "step": 21401
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3268701213664754e-06,
+      "loss": 0.3369,
+      "step": 21402
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.32603547891143e-06,
+      "loss": 0.2829,
+      "step": 21403
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3252009202817493e-06,
+      "loss": 0.3232,
+      "step": 21404
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.324366445487919e-06,
+      "loss": 0.2563,
+      "step": 21405
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3235320545404227e-06,
+      "loss": 0.3043,
+      "step": 21406
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.322697747449736e-06,
+      "loss": 0.3043,
+      "step": 21407
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3218635242263375e-06,
+      "loss": 0.323,
+      "step": 21408
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3210293848807117e-06,
+      "loss": 0.3061,
+      "step": 21409
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320195329423327e-06,
+      "loss": 0.3017,
+      "step": 21410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3193613578646633e-06,
+      "loss": 0.3377,
+      "step": 21411
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3185274702151993e-06,
+      "loss": 0.2554,
+      "step": 21412
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.317693666485403e-06,
+      "loss": 0.3528,
+      "step": 21413
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3168599466857466e-06,
+      "loss": 0.285,
+      "step": 21414
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3160263108267055e-06,
+      "loss": 0.311,
+      "step": 21415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.315192758918746e-06,
+      "loss": 0.3193,
+      "step": 21416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3143592909723387e-06,
+      "loss": 0.3162,
+      "step": 21417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3135259069979576e-06,
+      "loss": 0.3347,
+      "step": 21418
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312692607006062e-06,
+      "loss": 0.3083,
+      "step": 21419
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3118593910071216e-06,
+      "loss": 0.3472,
+      "step": 21420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.311026259011605e-06,
+      "loss": 0.3094,
+      "step": 21421
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3101932110299707e-06,
+      "loss": 0.3114,
+      "step": 21422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309360247072687e-06,
+      "loss": 0.3017,
+      "step": 21423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308527367150213e-06,
+      "loss": 0.2764,
+      "step": 21424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3076945712730068e-06,
+      "loss": 0.2991,
+      "step": 21425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3068618594515324e-06,
+      "loss": 0.3522,
+      "step": 21426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3060292316962505e-06,
+      "loss": 0.304,
+      "step": 21427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3051966880176136e-06,
+      "loss": 0.3553,
+      "step": 21428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304364228426081e-06,
+      "loss": 0.325,
+      "step": 21429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3035318529321116e-06,
+      "loss": 0.335,
+      "step": 21430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3026995615461543e-06,
+      "loss": 0.3502,
+      "step": 21431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301867354278665e-06,
+      "loss": 0.3104,
+      "step": 21432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3010352311401008e-06,
+      "loss": 0.3091,
+      "step": 21433
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3002031921409093e-06,
+      "loss": 0.3473,
+      "step": 21434
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2993712372915376e-06,
+      "loss": 0.3182,
+      "step": 21435
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.298539366602438e-06,
+      "loss": 0.2757,
+      "step": 21436
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.297707580084063e-06,
+      "loss": 0.3366,
+      "step": 21437
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2968758777468524e-06,
+      "loss": 0.3391,
+      "step": 21438
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2960442596012566e-06,
+      "loss": 0.3214,
+      "step": 21439
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2952127256577237e-06,
+      "loss": 0.3426,
+      "step": 21440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2943812759266914e-06,
+      "loss": 0.3083,
+      "step": 21441
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2935499104186053e-06,
+      "loss": 0.2975,
+      "step": 21442
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2927186291439117e-06,
+      "loss": 0.3177,
+      "step": 21443
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2918874321130466e-06,
+      "loss": 0.3114,
+      "step": 21444
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2910563193364497e-06,
+      "loss": 0.324,
+      "step": 21445
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2902252908245623e-06,
+      "loss": 0.3219,
+      "step": 21446
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.28939434658782e-06,
+      "loss": 0.3184,
+      "step": 21447
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.288563486636659e-06,
+      "loss": 0.3617,
+      "step": 21448
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2877327109815203e-06,
+      "loss": 0.2862,
+      "step": 21449
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2869020196328306e-06,
+      "loss": 0.3512,
+      "step": 21450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2860714126010286e-06,
+      "loss": 0.3406,
+      "step": 21451
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2852408898965483e-06,
+      "loss": 0.3403,
+      "step": 21452
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.284410451529816e-06,
+      "loss": 0.3258,
+      "step": 21453
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2835800975112673e-06,
+      "loss": 0.3259,
+      "step": 21454
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.282749827851325e-06,
+      "loss": 0.3357,
+      "step": 21455
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281919642560425e-06,
+      "loss": 0.2994,
+      "step": 21456
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281089541648987e-06,
+      "loss": 0.3055,
+      "step": 21457
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.280259525127444e-06,
+      "loss": 0.3123,
+      "step": 21458
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2794295930062147e-06,
+      "loss": 0.3248,
+      "step": 21459
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.278599745295725e-06,
+      "loss": 0.324,
+      "step": 21460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2777699820064035e-06,
+      "loss": 0.3494,
+      "step": 21461
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2769403031486637e-06,
+      "loss": 0.3106,
+      "step": 21462
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.27611070873293e-06,
+      "loss": 0.33,
+      "step": 21463
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2752811987696264e-06,
+      "loss": 0.2849,
+      "step": 21464
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.274451773269164e-06,
+      "loss": 0.3449,
+      "step": 21465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2736224322419663e-06,
+      "loss": 0.3137,
+      "step": 21466
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2727931756984487e-06,
+      "loss": 0.3377,
+      "step": 21467
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2719640036490218e-06,
+      "loss": 0.3235,
+      "step": 21468
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.271134916104104e-06,
+      "loss": 0.3399,
+      "step": 21469
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.270305913074112e-06,
+      "loss": 0.3063,
+      "step": 21470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2694769945694518e-06,
+      "loss": 0.3133,
+      "step": 21471
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.268648160600536e-06,
+      "loss": 0.2933,
+      "step": 21472
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.267819411177777e-06,
+      "loss": 0.2971,
+      "step": 21473
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.266990746311587e-06,
+      "loss": 0.2855,
+      "step": 21474
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2661621660123666e-06,
+      "loss": 0.3161,
+      "step": 21475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2653336702905305e-06,
+      "loss": 0.3006,
+      "step": 21476
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2645052591564796e-06,
+      "loss": 0.3303,
+      "step": 21477
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2636769326206176e-06,
+      "loss": 0.3141,
+      "step": 21478
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.26284869069335e-06,
+      "loss": 0.2859,
+      "step": 21479
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.262020533385084e-06,
+      "loss": 0.3574,
+      "step": 21480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.261192460706213e-06,
+      "loss": 0.2837,
+      "step": 21481
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2603644726671435e-06,
+      "loss": 0.3179,
+      "step": 21482
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2595365692782756e-06,
+      "loss": 0.3271,
+      "step": 21483
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2587087505500035e-06,
+      "loss": 0.3367,
+      "step": 21484
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2578810164927266e-06,
+      "loss": 0.32,
+      "step": 21485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2570533671168457e-06,
+      "loss": 0.3284,
+      "step": 21486
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.256225802432752e-06,
+      "loss": 0.3151,
+      "step": 21487
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255398322450837e-06,
+      "loss": 0.3138,
+      "step": 21488
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2545709271814987e-06,
+      "loss": 0.3546,
+      "step": 21489
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2537436166351245e-06,
+      "loss": 0.3331,
+      "step": 21490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2529163908221096e-06,
+      "loss": 0.3098,
+      "step": 21491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2520892497528446e-06,
+      "loss": 0.3316,
+      "step": 21492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2512621934377132e-06,
+      "loss": 0.2879,
+      "step": 21493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250435221887107e-06,
+      "loss": 0.3378,
+      "step": 21494
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2496083351114162e-06,
+      "loss": 0.3091,
+      "step": 21495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2487815331210194e-06,
+      "loss": 0.3013,
+      "step": 21496
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2479548159263074e-06,
+      "loss": 0.3274,
+      "step": 21497
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.247128183537661e-06,
+      "loss": 0.3254,
+      "step": 21498
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2463016359654607e-06,
+      "loss": 0.3437,
+      "step": 21499
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2454751732200894e-06,
+      "loss": 0.3193,
+      "step": 21500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.244648795311931e-06,
+      "loss": 0.3058,
+      "step": 21501
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.243822502251359e-06,
+      "loss": 0.3657,
+      "step": 21502
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242996294048755e-06,
+      "loss": 0.3299,
+      "step": 21503
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242170170714499e-06,
+      "loss": 0.3046,
+      "step": 21504
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.241344132258961e-06,
+      "loss": 0.2926,
+      "step": 21505
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24051817869252e-06,
+      "loss": 0.3385,
+      "step": 21506
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2396923100255515e-06,
+      "loss": 0.2636,
+      "step": 21507
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.238866526268426e-06,
+      "loss": 0.3192,
+      "step": 21508
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2380408274315135e-06,
+      "loss": 0.3299,
+      "step": 21509
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2372152135251855e-06,
+      "loss": 0.313,
+      "step": 21510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2363896845598175e-06,
+      "loss": 0.3304,
+      "step": 21511
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.235564240545771e-06,
+      "loss": 0.2715,
+      "step": 21512
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2347388814934154e-06,
+      "loss": 0.3119,
+      "step": 21513
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2339136074131217e-06,
+      "loss": 0.3423,
+      "step": 21514
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.233088418315249e-06,
+      "loss": 0.2994,
+      "step": 21515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2322633142101655e-06,
+      "loss": 0.3211,
+      "step": 21516
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2314382951082358e-06,
+      "loss": 0.3035,
+      "step": 21517
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.230613361019821e-06,
+      "loss": 0.3346,
+      "step": 21518
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2297885119552786e-06,
+      "loss": 0.3076,
+      "step": 21519
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2289637479249738e-06,
+      "loss": 0.3277,
+      "step": 21520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.228139068939261e-06,
+      "loss": 0.3142,
+      "step": 21521
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2273144750085005e-06,
+      "loss": 0.296,
+      "step": 21522
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2264899661430526e-06,
+      "loss": 0.2949,
+      "step": 21523
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.225665542353268e-06,
+      "loss": 0.3104,
+      "step": 21524
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2248412036495024e-06,
+      "loss": 0.343,
+      "step": 21525
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.224016950042114e-06,
+      "loss": 0.3114,
+      "step": 21526
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2231927815414487e-06,
+      "loss": 0.3296,
+      "step": 21527
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.222368698157866e-06,
+      "loss": 0.2896,
+      "step": 21528
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2215446999017107e-06,
+      "loss": 0.3185,
+      "step": 21529
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.220720786783331e-06,
+      "loss": 0.334,
+      "step": 21530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2198969588130793e-06,
+      "loss": 0.3049,
+      "step": 21531
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.219073216001303e-06,
+      "loss": 0.3347,
+      "step": 21532
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.218249558358344e-06,
+      "loss": 0.3215,
+      "step": 21533
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.21742598589455e-06,
+      "loss": 0.305,
+      "step": 21534
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2166024986202693e-06,
+      "loss": 0.3325,
+      "step": 21535
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.215779096545838e-06,
+      "loss": 0.3184,
+      "step": 21536
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214955779681601e-06,
+      "loss": 0.3167,
+      "step": 21537
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214132548037903e-06,
+      "loss": 0.3304,
+      "step": 21538
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2133094016250797e-06,
+      "loss": 0.3557,
+      "step": 21539
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2124863404534677e-06,
+      "loss": 0.3284,
+      "step": 21540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2116633645334094e-06,
+      "loss": 0.303,
+      "step": 21541
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2108404738752374e-06,
+      "loss": 0.3088,
+      "step": 21542
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2100176684892893e-06,
+      "loss": 0.3446,
+      "step": 21543
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.209194948385902e-06,
+      "loss": 0.34,
+      "step": 21544
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.208372313575404e-06,
+      "loss": 0.2971,
+      "step": 21545
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2075497640681297e-06,
+      "loss": 0.3037,
+      "step": 21546
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2067272998744113e-06,
+      "loss": 0.2995,
+      "step": 21547
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2059049210045812e-06,
+      "loss": 0.297,
+      "step": 21548
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.205082627468966e-06,
+      "loss": 0.3106,
+      "step": 21549
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2042604192778904e-06,
+      "loss": 0.3152,
+      "step": 21550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2034382964416876e-06,
+      "loss": 0.3422,
+      "step": 21551
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.202616258970677e-06,
+      "loss": 0.338,
+      "step": 21552
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2017943068751878e-06,
+      "loss": 0.3336,
+      "step": 21553
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2009724401655453e-06,
+      "loss": 0.3152,
+      "step": 21554
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.200150658852066e-06,
+      "loss": 0.3448,
+      "step": 21555
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1993289629450764e-06,
+      "loss": 0.2893,
+      "step": 21556
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.198507352454898e-06,
+      "loss": 0.3365,
+      "step": 21557
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1976858273918453e-06,
+      "loss": 0.2983,
+      "step": 21558
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196864387766244e-06,
+      "loss": 0.3364,
+      "step": 21559
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196043033588403e-06,
+      "loss": 0.318,
+      "step": 21560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1952217648686455e-06,
+      "loss": 0.3258,
+      "step": 21561
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.194400581617281e-06,
+      "loss": 0.3089,
+      "step": 21562
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1935794838446287e-06,
+      "loss": 0.3127,
+      "step": 21563
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1927584715609962e-06,
+      "loss": 0.2873,
+      "step": 21564
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.191937544776699e-06,
+      "loss": 0.2904,
+      "step": 21565
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1911167035020508e-06,
+      "loss": 0.3139,
+      "step": 21566
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.190295947747355e-06,
+      "loss": 0.3185,
+      "step": 21567
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.189475277522923e-06,
+      "loss": 0.3104,
+      "step": 21568
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.188654692839067e-06,
+      "loss": 0.2909,
+      "step": 21569
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1878341937060854e-06,
+      "loss": 0.3319,
+      "step": 21570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1870137801342917e-06,
+      "loss": 0.3012,
+      "step": 21571
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1861934521339867e-06,
+      "loss": 0.2972,
+      "step": 21572
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185373209715469e-06,
+      "loss": 0.287,
+      "step": 21573
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1845530528890465e-06,
+      "loss": 0.2701,
+      "step": 21574
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.183732981665022e-06,
+      "loss": 0.3111,
+      "step": 21575
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.182912996053691e-06,
+      "loss": 0.3315,
+      "step": 21576
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.182093096065354e-06,
+      "loss": 0.2958,
+      "step": 21577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.181273281710312e-06,
+      "loss": 0.301,
+      "step": 21578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1804535529988568e-06,
+      "loss": 0.2893,
+      "step": 21579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1796339099412866e-06,
+      "loss": 0.3006,
+      "step": 21580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1788143525479e-06,
+      "loss": 0.3802,
+      "step": 21581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1779948808289873e-06,
+      "loss": 0.3298,
+      "step": 21582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1771754947948376e-06,
+      "loss": 0.3005,
+      "step": 21583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1763561944557464e-06,
+      "loss": 0.3371,
+      "step": 21584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.175536979822006e-06,
+      "loss": 0.3036,
+      "step": 21585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1747178509039002e-06,
+      "loss": 0.3376,
+      "step": 21586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1738988077117205e-06,
+      "loss": 0.3184,
+      "step": 21587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.173079850255757e-06,
+      "loss": 0.2859,
+      "step": 21588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172260978546291e-06,
+      "loss": 0.3201,
+      "step": 21589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.171442192593608e-06,
+      "loss": 0.2826,
+      "step": 21590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.170623492407997e-06,
+      "loss": 0.3095,
+      "step": 21591
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.169804877999738e-06,
+      "loss": 0.3031,
+      "step": 21592
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1689863493791095e-06,
+      "loss": 0.2641,
+      "step": 21593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1681679065563976e-06,
+      "loss": 0.3047,
+      "step": 21594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167349549541876e-06,
+      "loss": 0.3052,
+      "step": 21595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.166531278345827e-06,
+      "loss": 0.3075,
+      "step": 21596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1657130929785307e-06,
+      "loss": 0.3289,
+      "step": 21597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1648949934502572e-06,
+      "loss": 0.3192,
+      "step": 21598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1640769797712865e-06,
+      "loss": 0.2982,
+      "step": 21599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163259051951895e-06,
+      "loss": 0.3262,
+      "step": 21600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.162441210002348e-06,
+      "loss": 0.3269,
+      "step": 21601
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.161623453932927e-06,
+      "loss": 0.3055,
+      "step": 21602
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1608057837538976e-06,
+      "loss": 0.3006,
+      "step": 21603
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1599881994755276e-06,
+      "loss": 0.2858,
+      "step": 21604
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1591707011080887e-06,
+      "loss": 0.2983,
+      "step": 21605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.158353288661853e-06,
+      "loss": 0.3095,
+      "step": 21606
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1575359621470793e-06,
+      "loss": 0.3058,
+      "step": 21607
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.156718721574037e-06,
+      "loss": 0.3271,
+      "step": 21608
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1559015669529948e-06,
+      "loss": 0.3216,
+      "step": 21609
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1550844982942075e-06,
+      "loss": 0.2959,
+      "step": 21610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.154267515607944e-06,
+      "loss": 0.33,
+      "step": 21611
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1534506189044658e-06,
+      "loss": 0.2842,
+      "step": 21612
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1526338081940323e-06,
+      "loss": 0.326,
+      "step": 21613
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1518170834868977e-06,
+      "loss": 0.3086,
+      "step": 21614
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.151000444793327e-06,
+      "loss": 0.3122,
+      "step": 21615
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1501838921235727e-06,
+      "loss": 0.3397,
+      "step": 21616
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1493674254878916e-06,
+      "loss": 0.2706,
+      "step": 21617
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1485510448965383e-06,
+      "loss": 0.2953,
+      "step": 21618
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.147734750359771e-06,
+      "loss": 0.3161,
+      "step": 21619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1469185418878368e-06,
+      "loss": 0.3226,
+      "step": 21620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1461024194909897e-06,
+      "loss": 0.3473,
+      "step": 21621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.145286383179482e-06,
+      "loss": 0.3267,
+      "step": 21622
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1444704329635623e-06,
+      "loss": 0.3437,
+      "step": 21623
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.143654568853474e-06,
+      "loss": 0.3153,
+      "step": 21624
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1428387908594717e-06,
+      "loss": 0.2982,
+      "step": 21625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1420230989917954e-06,
+      "loss": 0.3204,
+      "step": 21626
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1412074932606927e-06,
+      "loss": 0.3295,
+      "step": 21627
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.140391973676411e-06,
+      "loss": 0.3011,
+      "step": 21628
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1395765402491874e-06,
+      "loss": 0.2887,
+      "step": 21629
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.138761192989267e-06,
+      "loss": 0.3181,
+      "step": 21630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1379459319068927e-06,
+      "loss": 0.3106,
+      "step": 21631
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.137130757012299e-06,
+      "loss": 0.3448,
+      "step": 21632
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1363156683157316e-06,
+      "loss": 0.3108,
+      "step": 21633
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.135500665827422e-06,
+      "loss": 0.3297,
+      "step": 21634
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1346857495576066e-06,
+      "loss": 0.2937,
+      "step": 21635
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.133870919516522e-06,
+      "loss": 0.3039,
+      "step": 21636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1330561757144063e-06,
+      "loss": 0.3061,
+      "step": 21637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1322415181614876e-06,
+      "loss": 0.3457,
+      "step": 21638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.131426946867998e-06,
+      "loss": 0.291,
+      "step": 21639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.130612461844176e-06,
+      "loss": 0.3279,
+      "step": 21640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.129798063100242e-06,
+      "loss": 0.2822,
+      "step": 21641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1289837506464294e-06,
+      "loss": 0.3282,
+      "step": 21642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1281695244929676e-06,
+      "loss": 0.3349,
+      "step": 21643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1273553846500824e-06,
+      "loss": 0.3312,
+      "step": 21644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1265413311279957e-06,
+      "loss": 0.2974,
+      "step": 21645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1257273639369377e-06,
+      "loss": 0.3222,
+      "step": 21646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1249134830871265e-06,
+      "loss": 0.328,
+      "step": 21647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1240996885887866e-06,
+      "loss": 0.3339,
+      "step": 21648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.123285980452142e-06,
+      "loss": 0.3302,
+      "step": 21649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1224723586874084e-06,
+      "loss": 0.3096,
+      "step": 21650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121658823304806e-06,
+      "loss": 0.2755,
+      "step": 21651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1208453743145583e-06,
+      "loss": 0.3266,
+      "step": 21652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1200320117268745e-06,
+      "loss": 0.2862,
+      "step": 21653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1192187355519775e-06,
+      "loss": 0.3043,
+      "step": 21654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.118405545800075e-06,
+      "loss": 0.2831,
+      "step": 21655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1175924424813874e-06,
+      "loss": 0.3101,
+      "step": 21656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1167794256061213e-06,
+      "loss": 0.2773,
+      "step": 21657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.11596649518449e-06,
+      "loss": 0.3189,
+      "step": 21658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.11515365122671e-06,
+      "loss": 0.2973,
+      "step": 21659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114340893742982e-06,
+      "loss": 0.2824,
+      "step": 21660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.113528222743517e-06,
+      "loss": 0.3129,
+      "step": 21661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.112715638238527e-06,
+      "loss": 0.3187,
+      "step": 21662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1119031402382117e-06,
+      "loss": 0.3234,
+      "step": 21663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1110907287527815e-06,
+      "loss": 0.3088,
+      "step": 21664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110278403792434e-06,
+      "loss": 0.3328,
+      "step": 21665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.10946616536738e-06,
+      "loss": 0.3427,
+      "step": 21666
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1086540134878117e-06,
+      "loss": 0.2968,
+      "step": 21667
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.10784194816394e-06,
+      "loss": 0.3362,
+      "step": 21668
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1070299694059548e-06,
+      "loss": 0.3231,
+      "step": 21669
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1062180772240604e-06,
+      "loss": 0.2988,
+      "step": 21670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.105406271628455e-06,
+      "loss": 0.2999,
+      "step": 21671
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1045945526293307e-06,
+      "loss": 0.3129,
+      "step": 21672
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.103782920236885e-06,
+      "loss": 0.3016,
+      "step": 21673
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1029713744613156e-06,
+      "loss": 0.2991,
+      "step": 21674
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.102159915312808e-06,
+      "loss": 0.3298,
+      "step": 21675
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.101348542801562e-06,
+      "loss": 0.2834,
+      "step": 21676
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1005372569377644e-06,
+      "loss": 0.324,
+      "step": 21677
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099726057731601e-06,
+      "loss": 0.3139,
+      "step": 21678
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0989149451932667e-06,
+      "loss": 0.3068,
+      "step": 21679
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0981039193329487e-06,
+      "loss": 0.3086,
+      "step": 21680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09729298016083e-06,
+      "loss": 0.3086,
+      "step": 21681
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0964821276870983e-06,
+      "loss": 0.3113,
+      "step": 21682
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0956713619219404e-06,
+      "loss": 0.327,
+      "step": 21683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0948606828755336e-06,
+      "loss": 0.329,
+      "step": 21684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0940500905580627e-06,
+      "loss": 0.2633,
+      "step": 21685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0932395849797138e-06,
+      "loss": 0.3175,
+      "step": 21686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0924291661506613e-06,
+      "loss": 0.3064,
+      "step": 21687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0916188340810828e-06,
+      "loss": 0.3073,
+      "step": 21688
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.090808588781162e-06,
+      "loss": 0.298,
+      "step": 21689
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0899984302610696e-06,
+      "loss": 0.3249,
+      "step": 21690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0891883585309833e-06,
+      "loss": 0.3184,
+      "step": 21691
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0883783736010776e-06,
+      "loss": 0.2796,
+      "step": 21692
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0875684754815306e-06,
+      "loss": 0.3282,
+      "step": 21693
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0867586641825074e-06,
+      "loss": 0.3341,
+      "step": 21694
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0859489397141827e-06,
+      "loss": 0.3237,
+      "step": 21695
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0851393020867294e-06,
+      "loss": 0.3559,
+      "step": 21696
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.084329751310314e-06,
+      "loss": 0.3028,
+      "step": 21697
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.083520287395101e-06,
+      "loss": 0.3063,
+      "step": 21698
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0827109103512643e-06,
+      "loss": 0.3272,
+      "step": 21699
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0819016201889617e-06,
+      "loss": 0.2678,
+      "step": 21700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0810924169183633e-06,
+      "loss": 0.3175,
+      "step": 21701
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0802833005496337e-06,
+      "loss": 0.319,
+      "step": 21702
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0794742710929304e-06,
+      "loss": 0.3315,
+      "step": 21703
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078665328558419e-06,
+      "loss": 0.3057,
+      "step": 21704
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.077856472956261e-06,
+      "loss": 0.2999,
+      "step": 21705
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0770477042966095e-06,
+      "loss": 0.3461,
+      "step": 21706
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076239022589631e-06,
+      "loss": 0.3283,
+      "step": 21707
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.075430427845477e-06,
+      "loss": 0.327,
+      "step": 21708
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.074621920074303e-06,
+      "loss": 0.2994,
+      "step": 21709
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0738134992862646e-06,
+      "loss": 0.3315,
+      "step": 21710
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0730051654915193e-06,
+      "loss": 0.3093,
+      "step": 21711
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0721969187002155e-06,
+      "loss": 0.3097,
+      "step": 21712
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0713887589225045e-06,
+      "loss": 0.3376,
+      "step": 21713
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.070580686168543e-06,
+      "loss": 0.3087,
+      "step": 21714
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0697727004484725e-06,
+      "loss": 0.3358,
+      "step": 21715
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.068964801772446e-06,
+      "loss": 0.3018,
+      "step": 21716
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0681569901506115e-06,
+      "loss": 0.3203,
+      "step": 21717
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0673492655931146e-06,
+      "loss": 0.3349,
+      "step": 21718
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0665416281100957e-06,
+      "loss": 0.3064,
+      "step": 21719
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.065734077711705e-06,
+      "loss": 0.3003,
+      "step": 21720
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0649266144080804e-06,
+      "loss": 0.3155,
+      "step": 21721
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0641192382093642e-06,
+      "loss": 0.32,
+      "step": 21722
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0633119491257035e-06,
+      "loss": 0.3586,
+      "step": 21723
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0625047471672296e-06,
+      "loss": 0.3339,
+      "step": 21724
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.061697632344085e-06,
+      "loss": 0.3015,
+      "step": 21725
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0608906046664087e-06,
+      "loss": 0.3389,
+      "step": 21726
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0600836641443333e-06,
+      "loss": 0.3078,
+      "step": 21727
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.059276810787999e-06,
+      "loss": 0.2874,
+      "step": 21728
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.058470044607532e-06,
+      "loss": 0.3126,
+      "step": 21729
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.057663365613074e-06,
+      "loss": 0.304,
+      "step": 21730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0568567738147505e-06,
+      "loss": 0.2766,
+      "step": 21731
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0560502692226936e-06,
+      "loss": 0.3203,
+      "step": 21732
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0552438518470384e-06,
+      "loss": 0.3129,
+      "step": 21733
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.054437521697905e-06,
+      "loss": 0.3184,
+      "step": 21734
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.053631278785426e-06,
+      "loss": 0.3341,
+      "step": 21735
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0528251231197302e-06,
+      "loss": 0.3257,
+      "step": 21736
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0520190547109372e-06,
+      "loss": 0.3436,
+      "step": 21737
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.051213073569177e-06,
+      "loss": 0.2836,
+      "step": 21738
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0504071797045666e-06,
+      "loss": 0.3338,
+      "step": 21739
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0496013731272343e-06,
+      "loss": 0.2943,
+      "step": 21740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0487956538472953e-06,
+      "loss": 0.2939,
+      "step": 21741
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.047990021874875e-06,
+      "loss": 0.3283,
+      "step": 21742
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0471844772200856e-06,
+      "loss": 0.3091,
+      "step": 21743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0463790198930497e-06,
+      "loss": 0.3017,
+      "step": 21744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0455736499038847e-06,
+      "loss": 0.3165,
+      "step": 21745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0447683672627015e-06,
+      "loss": 0.3001,
+      "step": 21746
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.043963171979618e-06,
+      "loss": 0.3204,
+      "step": 21747
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0431580640647483e-06,
+      "loss": 0.2922,
+      "step": 21748
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.042353043528201e-06,
+      "loss": 0.3187,
+      "step": 21749
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0415481103800915e-06,
+      "loss": 0.3388,
+      "step": 21750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.040743264630528e-06,
+      "loss": 0.2952,
+      "step": 21751
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.039938506289616e-06,
+      "loss": 0.3332,
+      "step": 21752
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.039133835367466e-06,
+      "loss": 0.3313,
+      "step": 21753
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038329251874187e-06,
+      "loss": 0.3224,
+      "step": 21754
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0375247558198807e-06,
+      "loss": 0.3077,
+      "step": 21755
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0367203472146524e-06,
+      "loss": 0.3247,
+      "step": 21756
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0359160260686104e-06,
+      "loss": 0.3027,
+      "step": 21757
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0351117923918493e-06,
+      "loss": 0.3232,
+      "step": 21758
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0343076461944744e-06,
+      "loss": 0.3324,
+      "step": 21759
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0335035874865893e-06,
+      "loss": 0.3056,
+      "step": 21760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0326996162782885e-06,
+      "loss": 0.3397,
+      "step": 21761
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031895732579668e-06,
+      "loss": 0.3263,
+      "step": 21762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031091936400831e-06,
+      "loss": 0.3139,
+      "step": 21763
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.030288227751865e-06,
+      "loss": 0.321,
+      "step": 21764
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0294846066428707e-06,
+      "loss": 0.3247,
+      "step": 21765
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.028681073083939e-06,
+      "loss": 0.2824,
+      "step": 21766
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027877627085167e-06,
+      "loss": 0.3169,
+      "step": 21767
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027074268656639e-06,
+      "loss": 0.2927,
+      "step": 21768
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0262709978084526e-06,
+      "loss": 0.3293,
+      "step": 21769
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02546781455069e-06,
+      "loss": 0.3219,
+      "step": 21770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0246647188934454e-06,
+      "loss": 0.3297,
+      "step": 21771
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.023861710846798e-06,
+      "loss": 0.3245,
+      "step": 21772
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.023058790420843e-06,
+      "loss": 0.2626,
+      "step": 21773
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.022255957625657e-06,
+      "loss": 0.3154,
+      "step": 21774
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.021453212471326e-06,
+      "loss": 0.3386,
+      "step": 21775
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0206505549679377e-06,
+      "loss": 0.2781,
+      "step": 21776
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0198479851255656e-06,
+      "loss": 0.3197,
+      "step": 21777
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0190455029542943e-06,
+      "loss": 0.3153,
+      "step": 21778
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.018243108464205e-06,
+      "loss": 0.3142,
+      "step": 21779
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.01744080166537e-06,
+      "loss": 0.3071,
+      "step": 21780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0166385825678725e-06,
+      "loss": 0.36,
+      "step": 21781
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.015836451181786e-06,
+      "loss": 0.3158,
+      "step": 21782
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.015034407517182e-06,
+      "loss": 0.3056,
+      "step": 21783
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.014232451584137e-06,
+      "loss": 0.298,
+      "step": 21784
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0134305833927267e-06,
+      "loss": 0.3282,
+      "step": 21785
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012628802953016e-06,
+      "loss": 0.3239,
+      "step": 21786
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0118271102750794e-06,
+      "loss": 0.3027,
+      "step": 21787
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0110255053689884e-06,
+      "loss": 0.3257,
+      "step": 21788
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010223988244806e-06,
+      "loss": 0.3215,
+      "step": 21789
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0094225589126016e-06,
+      "loss": 0.3187,
+      "step": 21790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0086212173824435e-06,
+      "loss": 0.3184,
+      "step": 21791
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0078199636643956e-06,
+      "loss": 0.3185,
+      "step": 21792
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.007018797768516e-06,
+      "loss": 0.305,
+      "step": 21793
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.006217719704877e-06,
+      "loss": 0.3107,
+      "step": 21794
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0054167294835314e-06,
+      "loss": 0.3226,
+      "step": 21795
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004615827114542e-06,
+      "loss": 0.3302,
+      "step": 21796
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0038150126079746e-06,
+      "loss": 0.3256,
+      "step": 21797
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0030142859738786e-06,
+      "loss": 0.2831,
+      "step": 21798
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002213647222315e-06,
+      "loss": 0.3346,
+      "step": 21799
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0014130963633393e-06,
+      "loss": 0.272,
+      "step": 21800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.000612633407011e-06,
+      "loss": 0.3135,
+      "step": 21801
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9998122583633804e-06,
+      "loss": 0.3137,
+      "step": 21802
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9990119712424957e-06,
+      "loss": 0.3163,
+      "step": 21803
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9982117720544167e-06,
+      "loss": 0.2785,
+      "step": 21804
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9974116608091865e-06,
+      "loss": 0.3181,
+      "step": 21805
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9966116375168587e-06,
+      "loss": 0.323,
+      "step": 21806
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.995811702187483e-06,
+      "loss": 0.3012,
+      "step": 21807
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9950118548311012e-06,
+      "loss": 0.3095,
+      "step": 21808
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9942120954577646e-06,
+      "loss": 0.3215,
+      "step": 21809
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9934124240775185e-06,
+      "loss": 0.3394,
+      "step": 21810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9926128407004017e-06,
+      "loss": 0.3278,
+      "step": 21811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.991813345336463e-06,
+      "loss": 0.3224,
+      "step": 21812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9910139379957413e-06,
+      "loss": 0.3061,
+      "step": 21813
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990214618688274e-06,
+      "loss": 0.3001,
+      "step": 21814
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.989415387424104e-06,
+      "loss": 0.3324,
+      "step": 21815
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9886162442132716e-06,
+      "loss": 0.3447,
+      "step": 21816
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9878171890658102e-06,
+      "loss": 0.2928,
+      "step": 21817
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9870182219917564e-06,
+      "loss": 0.3176,
+      "step": 21818
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9862193430011497e-06,
+      "loss": 0.2988,
+      "step": 21819
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.985420552104018e-06,
+      "loss": 0.3257,
+      "step": 21820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9846218493103975e-06,
+      "loss": 0.3196,
+      "step": 21821
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.983823234630322e-06,
+      "loss": 0.3403,
+      "step": 21822
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9830247080738197e-06,
+      "loss": 0.329,
+      "step": 21823
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9822262696509176e-06,
+      "loss": 0.323,
+      "step": 21824
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.981427919371649e-06,
+      "loss": 0.286,
+      "step": 21825
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9806296572460358e-06,
+      "loss": 0.3075,
+      "step": 21826
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.979831483284107e-06,
+      "loss": 0.2918,
+      "step": 21827
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9790333974958917e-06,
+      "loss": 0.3332,
+      "step": 21828
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.978235399891407e-06,
+      "loss": 0.314,
+      "step": 21829
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.977437490480678e-06,
+      "loss": 0.3215,
+      "step": 21830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.976639669273731e-06,
+      "loss": 0.3163,
+      "step": 21831
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97584193628058e-06,
+      "loss": 0.3185,
+      "step": 21832
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975044291511251e-06,
+      "loss": 0.3388,
+      "step": 21833
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9742467349757563e-06,
+      "loss": 0.3203,
+      "step": 21834
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9734492666841184e-06,
+      "loss": 0.3161,
+      "step": 21835
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9726518866463493e-06,
+      "loss": 0.3024,
+      "step": 21836
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9718545948724652e-06,
+      "loss": 0.3009,
+      "step": 21837
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.971057391372485e-06,
+      "loss": 0.3512,
+      "step": 21838
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9702602761564136e-06,
+      "loss": 0.2943,
+      "step": 21839
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.969463249234267e-06,
+      "loss": 0.3529,
+      "step": 21840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9686663106160585e-06,
+      "loss": 0.3486,
+      "step": 21841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9678694603117917e-06,
+      "loss": 0.298,
+      "step": 21842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9670726983314823e-06,
+      "loss": 0.3134,
+      "step": 21843
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.966276024685131e-06,
+      "loss": 0.2915,
+      "step": 21844
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965479439382749e-06,
+      "loss": 0.3408,
+      "step": 21845
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.964682942434336e-06,
+      "loss": 0.2786,
+      "step": 21846
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963886533849902e-06,
+      "loss": 0.2853,
+      "step": 21847
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9630902136394455e-06,
+      "loss": 0.3148,
+      "step": 21848
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.962293981812969e-06,
+      "loss": 0.2934,
+      "step": 21849
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9614978383804782e-06,
+      "loss": 0.2883,
+      "step": 21850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9607017833519647e-06,
+      "loss": 0.295,
+      "step": 21851
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9599058167374304e-06,
+      "loss": 0.2714,
+      "step": 21852
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.959109938546878e-06,
+      "loss": 0.3233,
+      "step": 21853
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9583141487902955e-06,
+      "loss": 0.3066,
+      "step": 21854
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9575184474776852e-06,
+      "loss": 0.2997,
+      "step": 21855
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9567228346190367e-06,
+      "loss": 0.3218,
+      "step": 21856
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9559273102243414e-06,
+      "loss": 0.3309,
+      "step": 21857
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.955131874303593e-06,
+      "loss": 0.2902,
+      "step": 21858
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9543365268667866e-06,
+      "loss": 0.2687,
+      "step": 21859
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9535412679239038e-06,
+      "loss": 0.3106,
+      "step": 21860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9527460974849386e-06,
+      "loss": 0.3159,
+      "step": 21861
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9519510155598796e-06,
+      "loss": 0.3036,
+      "step": 21862
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.951156022158708e-06,
+      "loss": 0.3444,
+      "step": 21863
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9503611172914116e-06,
+      "loss": 0.3049,
+      "step": 21864
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9495663009679764e-06,
+      "loss": 0.2696,
+      "step": 21865
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9487715731983846e-06,
+      "loss": 0.3082,
+      "step": 21866
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.947976933992612e-06,
+      "loss": 0.3173,
+      "step": 21867
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9471823833606485e-06,
+      "loss": 0.287,
+      "step": 21868
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9463879213124658e-06,
+      "loss": 0.3306,
+      "step": 21869
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9455935478580467e-06,
+      "loss": 0.3023,
+      "step": 21870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9447992630073697e-06,
+      "loss": 0.3211,
+      "step": 21871
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9440050667704066e-06,
+      "loss": 0.3145,
+      "step": 21872
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9432109591571344e-06,
+      "loss": 0.3368,
+      "step": 21873
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9424169401775284e-06,
+      "loss": 0.3296,
+      "step": 21874
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.941623009841563e-06,
+      "loss": 0.3256,
+      "step": 21875
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9408291681592094e-06,
+      "loss": 0.3507,
+      "step": 21876
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.940035415140433e-06,
+      "loss": 0.3223,
+      "step": 21877
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.93924175079521e-06,
+      "loss": 0.3401,
+      "step": 21878
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9384481751335035e-06,
+      "loss": 0.3151,
+      "step": 21879
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9376546881652835e-06,
+      "loss": 0.3348,
+      "step": 21880
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9368612899005187e-06,
+      "loss": 0.3122,
+      "step": 21881
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9360679803491686e-06,
+      "loss": 0.3208,
+      "step": 21882
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9352747595212007e-06,
+      "loss": 0.2918,
+      "step": 21883
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9344816274265797e-06,
+      "loss": 0.3196,
+      "step": 21884
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9336885840752627e-06,
+      "loss": 0.333,
+      "step": 21885
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.932895629477216e-06,
+      "loss": 0.3081,
+      "step": 21886
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.932102763642396e-06,
+      "loss": 0.3186,
+      "step": 21887
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.931309986580758e-06,
+      "loss": 0.3292,
+      "step": 21888
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9305172983022624e-06,
+      "loss": 0.2781,
+      "step": 21889
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9297246988168694e-06,
+      "loss": 0.34,
+      "step": 21890
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.2896,
+      "step": 21891
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.92813976626519e-06,
+      "loss": 0.3185,
+      "step": 21892
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.927347433218819e-06,
+      "loss": 0.2994,
+      "step": 21893
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.926555189005356e-06,
+      "loss": 0.3297,
+      "step": 21894
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9257630336347564e-06,
+      "loss": 0.2879,
+      "step": 21895
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9249709671169725e-06,
+      "loss": 0.3489,
+      "step": 21896
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.92417898946195e-06,
+      "loss": 0.3188,
+      "step": 21897
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.923387100679632e-06,
+      "loss": 0.2881,
+      "step": 21898
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9225953007799713e-06,
+      "loss": 0.3079,
+      "step": 21899
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921803589772908e-06,
+      "loss": 0.3211,
+      "step": 21900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921011967668387e-06,
+      "loss": 0.273,
+      "step": 21901
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.920220434476355e-06,
+      "loss": 0.3506,
+      "step": 21902
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9194289902067483e-06,
+      "loss": 0.3065,
+      "step": 21903
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.91863763486951e-06,
+      "loss": 0.3228,
+      "step": 21904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9178463684745827e-06,
+      "loss": 0.3046,
+      "step": 21905
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.917055191031899e-06,
+      "loss": 0.3199,
+      "step": 21906
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9162641025514014e-06,
+      "loss": 0.2905,
+      "step": 21907
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9154731030430227e-06,
+      "loss": 0.3282,
+      "step": 21908
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.914682192516697e-06,
+      "loss": 0.3049,
+      "step": 21909
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913891370982359e-06,
+      "loss": 0.321,
+      "step": 21910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9131006384499415e-06,
+      "loss": 0.2775,
+      "step": 21911
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9123099949293807e-06,
+      "loss": 0.3188,
+      "step": 21912
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9115194404306013e-06,
+      "loss": 0.3022,
+      "step": 21913
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9107289749635336e-06,
+      "loss": 0.3066,
+      "step": 21914
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.90993859853811e-06,
+      "loss": 0.3129,
+      "step": 21915
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9091483111642514e-06,
+      "loss": 0.3211,
+      "step": 21916
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.908358112851891e-06,
+      "loss": 0.3035,
+      "step": 21917
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.90756800361095e-06,
+      "loss": 0.3005,
+      "step": 21918
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9067779834513476e-06,
+      "loss": 0.3547,
+      "step": 21919
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.905988052383012e-06,
+      "loss": 0.308,
+      "step": 21920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.905198210415867e-06,
+      "loss": 0.2959,
+      "step": 21921
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9044084575598263e-06,
+      "loss": 0.3287,
+      "step": 21922
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9036187938248126e-06,
+      "loss": 0.3355,
+      "step": 21923
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9028292192207484e-06,
+      "loss": 0.3284,
+      "step": 21924
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9020397337575434e-06,
+      "loss": 0.3345,
+      "step": 21925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.901250337445116e-06,
+      "loss": 0.3105,
+      "step": 21926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9004610302933854e-06,
+      "loss": 0.2684,
+      "step": 21927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8996718123122623e-06,
+      "loss": 0.2874,
+      "step": 21928
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8988826835116556e-06,
+      "loss": 0.3084,
+      "step": 21929
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.898093643901483e-06,
+      "loss": 0.2787,
+      "step": 21930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.897304693491648e-06,
+      "loss": 0.349,
+      "step": 21931
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896515832292065e-06,
+      "loss": 0.3086,
+      "step": 21932
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895727060312644e-06,
+      "loss": 0.2828,
+      "step": 21933
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.894938377563287e-06,
+      "loss": 0.3211,
+      "step": 21934
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8941497840539002e-06,
+      "loss": 0.2823,
+      "step": 21935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.893361279794393e-06,
+      "loss": 0.3248,
+      "step": 21936
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8925728647946637e-06,
+      "loss": 0.3349,
+      "step": 21937
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8917845390646205e-06,
+      "loss": 0.2887,
+      "step": 21938
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8909963026141585e-06,
+      "loss": 0.3318,
+      "step": 21939
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8902081554531846e-06,
+      "loss": 0.3218,
+      "step": 21940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889420097591591e-06,
+      "loss": 0.3116,
+      "step": 21941
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8886321290392817e-06,
+      "loss": 0.3262,
+      "step": 21942
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.887844249806149e-06,
+      "loss": 0.3464,
+      "step": 21943
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.88705645990209e-06,
+      "loss": 0.2974,
+      "step": 21944
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.886268759337003e-06,
+      "loss": 0.3239,
+      "step": 21945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8854811481207767e-06,
+      "loss": 0.2881,
+      "step": 21946
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8846936262633053e-06,
+      "loss": 0.3088,
+      "step": 21947
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8839061937744837e-06,
+      "loss": 0.3195,
+      "step": 21948
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.883118850664196e-06,
+      "loss": 0.3163,
+      "step": 21949
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882331596942336e-06,
+      "loss": 0.3009,
+      "step": 21950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.881544432618788e-06,
+      "loss": 0.3535,
+      "step": 21951
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8807573577034433e-06,
+      "loss": 0.2717,
+      "step": 21952
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.879970372206181e-06,
+      "loss": 0.2935,
+      "step": 21953
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8791834761368908e-06,
+      "loss": 0.2873,
+      "step": 21954
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8783966695054575e-06,
+      "loss": 0.3173,
+      "step": 21955
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877609952321758e-06,
+      "loss": 0.3153,
+      "step": 21956
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.876823324595677e-06,
+      "loss": 0.3508,
+      "step": 21957
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8760367863370953e-06,
+      "loss": 0.311,
+      "step": 21958
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8752503375558895e-06,
+      "loss": 0.3163,
+      "step": 21959
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87446397826194e-06,
+      "loss": 0.2775,
+      "step": 21960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8736777084651225e-06,
+      "loss": 0.3034,
+      "step": 21961
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872891528175309e-06,
+      "loss": 0.2862,
+      "step": 21962
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872105437402377e-06,
+      "loss": 0.3176,
+      "step": 21963
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8713194361562036e-06,
+      "loss": 0.3289,
+      "step": 21964
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8705335244466535e-06,
+      "loss": 0.2988,
+      "step": 21965
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8697477022836017e-06,
+      "loss": 0.2504,
+      "step": 21966
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868961969676922e-06,
+      "loss": 0.2916,
+      "step": 21967
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868176326636476e-06,
+      "loss": 0.3049,
+      "step": 21968
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867390773172134e-06,
+      "loss": 0.2623,
+      "step": 21969
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8666053092937674e-06,
+      "loss": 0.2856,
+      "step": 21970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.865819935011237e-06,
+      "loss": 0.3211,
+      "step": 21971
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.865034650334404e-06,
+      "loss": 0.2653,
+      "step": 21972
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8642494552731393e-06,
+      "loss": 0.2935,
+      "step": 21973
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8634643498372984e-06,
+      "loss": 0.3334,
+      "step": 21974
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862679334036744e-06,
+      "loss": 0.2985,
+      "step": 21975
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8618944078813414e-06,
+      "loss": 0.2721,
+      "step": 21976
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.86110957138094e-06,
+      "loss": 0.3053,
+      "step": 21977
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.860324824545404e-06,
+      "loss": 0.3427,
+      "step": 21978
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8595401673845902e-06,
+      "loss": 0.3287,
+      "step": 21979
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8587555999083484e-06,
+      "loss": 0.3069,
+      "step": 21980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8579711221265396e-06,
+      "loss": 0.3245,
+      "step": 21981
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8571867340490134e-06,
+      "loss": 0.3164,
+      "step": 21982
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.856402435685619e-06,
+      "loss": 0.3186,
+      "step": 21983
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8556182270462094e-06,
+      "loss": 0.3103,
+      "step": 21984
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854834108140636e-06,
+      "loss": 0.3326,
+      "step": 21985
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8540500789787494e-06,
+      "loss": 0.3594,
+      "step": 21986
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853266139570391e-06,
+      "loss": 0.2934,
+      "step": 21987
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.852482289925409e-06,
+      "loss": 0.3064,
+      "step": 21988
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8516985300536537e-06,
+      "loss": 0.3266,
+      "step": 21989
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8509148599649627e-06,
+      "loss": 0.3125,
+      "step": 21990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8501312796691837e-06,
+      "loss": 0.3411,
+      "step": 21991
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8493477891761558e-06,
+      "loss": 0.2976,
+      "step": 21992
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.848564388495718e-06,
+      "loss": 0.2958,
+      "step": 21993
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.847781077637711e-06,
+      "loss": 0.3278,
+      "step": 21994
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.846997856611978e-06,
+      "loss": 0.3121,
+      "step": 21995
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.84621472542835e-06,
+      "loss": 0.3613,
+      "step": 21996
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8454316840966656e-06,
+      "loss": 0.3303,
+      "step": 21997
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8446487326267625e-06,
+      "loss": 0.3322,
+      "step": 21998
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8438658710284696e-06,
+      "loss": 0.2712,
+      "step": 21999
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.843083099311621e-06,
+      "loss": 0.3185,
+      "step": 22000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8423004174860538e-06,
+      "loss": 0.3051,
+      "step": 22001
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8415178255615926e-06,
+      "loss": 0.2857,
+      "step": 22002
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8407353235480673e-06,
+      "loss": 0.3118,
+      "step": 22003
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8399529114553093e-06,
+      "loss": 0.3676,
+      "step": 22004
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.839170589293141e-06,
+      "loss": 0.2887,
+      "step": 22005
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.838388357071391e-06,
+      "loss": 0.309,
+      "step": 22006
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.837606214799887e-06,
+      "loss": 0.3021,
+      "step": 22007
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8368241624884473e-06,
+      "loss": 0.326,
+      "step": 22008
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8360422001468978e-06,
+      "loss": 0.334,
+      "step": 22009
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8352603277850623e-06,
+      "loss": 0.2767,
+      "step": 22010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.834478545412754e-06,
+      "loss": 0.3143,
+      "step": 22011
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8336968530398012e-06,
+      "loss": 0.3156,
+      "step": 22012
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8329152506760137e-06,
+      "loss": 0.3337,
+      "step": 22013
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8321337383312154e-06,
+      "loss": 0.3246,
+      "step": 22014
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8313523160152156e-06,
+      "loss": 0.3025,
+      "step": 22015
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8305709837378347e-06,
+      "loss": 0.3093,
+      "step": 22016
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8297897415088817e-06,
+      "loss": 0.3224,
+      "step": 22017
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8290085893381703e-06,
+      "loss": 0.305,
+      "step": 22018
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.828227527235513e-06,
+      "loss": 0.3227,
+      "step": 22019
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.827446555210722e-06,
+      "loss": 0.309,
+      "step": 22020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826665673273602e-06,
+      "loss": 0.3447,
+      "step": 22021
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8258848814339655e-06,
+      "loss": 0.3041,
+      "step": 22022
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8251041797016167e-06,
+      "loss": 0.3343,
+      "step": 22023
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8243235680863577e-06,
+      "loss": 0.3239,
+      "step": 22024
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8235430465979975e-06,
+      "loss": 0.3168,
+      "step": 22025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8227626152463404e-06,
+      "loss": 0.2896,
+      "step": 22026
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.821982274041185e-06,
+      "loss": 0.3071,
+      "step": 22027
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8212020229923344e-06,
+      "loss": 0.3107,
+      "step": 22028
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8204218621095904e-06,
+      "loss": 0.3091,
+      "step": 22029
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8196417914027474e-06,
+      "loss": 0.3314,
+      "step": 22030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.818861810881606e-06,
+      "loss": 0.3279,
+      "step": 22031
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.818081920555964e-06,
+      "loss": 0.3102,
+      "step": 22032
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8173021204356165e-06,
+      "loss": 0.3214,
+      "step": 22033
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8165224105303534e-06,
+      "loss": 0.3182,
+      "step": 22034
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8157427908499747e-06,
+      "loss": 0.3411,
+      "step": 22035
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8149632614042653e-06,
+      "loss": 0.302,
+      "step": 22036
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8141838222030195e-06,
+      "loss": 0.2742,
+      "step": 22037
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8134044732560306e-06,
+      "loss": 0.3135,
+      "step": 22038
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8126252145730815e-06,
+      "loss": 0.3089,
+      "step": 22039
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8118460461639617e-06,
+      "loss": 0.3087,
+      "step": 22040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8110669680384618e-06,
+      "loss": 0.3048,
+      "step": 22041
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8102879802063597e-06,
+      "loss": 0.296,
+      "step": 22042
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8095090826774464e-06,
+      "loss": 0.3125,
+      "step": 22043
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.808730275461499e-06,
+      "loss": 0.3294,
+      "step": 22044
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8079515585683048e-06,
+      "loss": 0.3384,
+      "step": 22045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8071729320076378e-06,
+      "loss": 0.3478,
+      "step": 22046
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.806394395789287e-06,
+      "loss": 0.3519,
+      "step": 22047
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8056159499230207e-06,
+      "loss": 0.3196,
+      "step": 22048
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804837594418621e-06,
+      "loss": 0.2916,
+      "step": 22049
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8040593292858674e-06,
+      "loss": 0.3083,
+      "step": 22050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8032811545345294e-06,
+      "loss": 0.3382,
+      "step": 22051
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8025030701743827e-06,
+      "loss": 0.3315,
+      "step": 22052
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8017250762152037e-06,
+      "loss": 0.3139,
+      "step": 22053
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.800947172666757e-06,
+      "loss": 0.2749,
+      "step": 22054
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.800169359538821e-06,
+      "loss": 0.3256,
+      "step": 22055
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.799391636841158e-06,
+      "loss": 0.3012,
+      "step": 22056
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.798614004583542e-06,
+      "loss": 0.3179,
+      "step": 22057
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7978364627757336e-06,
+      "loss": 0.3125,
+      "step": 22058
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7970590114275042e-06,
+      "loss": 0.2847,
+      "step": 22059
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.796281650548619e-06,
+      "loss": 0.2688,
+      "step": 22060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7955043801488368e-06,
+      "loss": 0.3534,
+      "step": 22061
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.794727200237923e-06,
+      "loss": 0.2951,
+      "step": 22062
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7939501108256427e-06,
+      "loss": 0.3206,
+      "step": 22063
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7931731119217486e-06,
+      "loss": 0.3106,
+      "step": 22064
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7923962035360084e-06,
+      "loss": 0.3365,
+      "step": 22065
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.791619385678175e-06,
+      "loss": 0.3262,
+      "step": 22066
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.790842658358002e-06,
+      "loss": 0.3326,
+      "step": 22067
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7900660215852494e-06,
+      "loss": 0.3366,
+      "step": 22068
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.789289475369674e-06,
+      "loss": 0.3031,
+      "step": 22069
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.788513019721024e-06,
+      "loss": 0.3116,
+      "step": 22070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.787736654649055e-06,
+      "loss": 0.3038,
+      "step": 22071
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.78696038016352e-06,
+      "loss": 0.3057,
+      "step": 22072
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7861841962741633e-06,
+      "loss": 0.3403,
+      "step": 22073
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7854081029907376e-06,
+      "loss": 0.2816,
+      "step": 22074
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7846321003229927e-06,
+      "loss": 0.327,
+      "step": 22075
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7838561882806737e-06,
+      "loss": 0.304,
+      "step": 22076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7830803668735207e-06,
+      "loss": 0.3125,
+      "step": 22077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.782304636111286e-06,
+      "loss": 0.3098,
+      "step": 22078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.781528996003707e-06,
+      "loss": 0.3403,
+      "step": 22079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.780753446560528e-06,
+      "loss": 0.32,
+      "step": 22080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7799779877914925e-06,
+      "loss": 0.3327,
+      "step": 22081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7792026197063347e-06,
+      "loss": 0.3477,
+      "step": 22082
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7784273423147966e-06,
+      "loss": 0.3334,
+      "step": 22083
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7776521556266188e-06,
+      "loss": 0.2895,
+      "step": 22084
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7768770596515313e-06,
+      "loss": 0.3137,
+      "step": 22085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7761020543992757e-06,
+      "loss": 0.333,
+      "step": 22086
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.775327139879582e-06,
+      "loss": 0.3353,
+      "step": 22087
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7745523161021813e-06,
+      "loss": 0.2872,
+      "step": 22088
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7737775830768087e-06,
+      "loss": 0.3017,
+      "step": 22089
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7730029408131974e-06,
+      "loss": 0.2991,
+      "step": 22090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.772228389321071e-06,
+      "loss": 0.3076,
+      "step": 22091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7714539286101604e-06,
+      "loss": 0.3133,
+      "step": 22092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7706795586901935e-06,
+      "loss": 0.3369,
+      "step": 22093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.769905279570899e-06,
+      "loss": 0.3611,
+      "step": 22094
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7691310912619963e-06,
+      "loss": 0.319,
+      "step": 22095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7683569937732147e-06,
+      "loss": 0.309,
+      "step": 22096
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7675829871142747e-06,
+      "loss": 0.3418,
+      "step": 22097
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.766809071294895e-06,
+      "loss": 0.3012,
+      "step": 22098
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7660352463247973e-06,
+      "loss": 0.3005,
+      "step": 22099
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.765261512213706e-06,
+      "loss": 0.3037,
+      "step": 22100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7644878689713316e-06,
+      "loss": 0.2899,
+      "step": 22101
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7637143166073956e-06,
+      "loss": 0.2709,
+      "step": 22102
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.762940855131615e-06,
+      "loss": 0.3245,
+      "step": 22103
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7621674845537006e-06,
+      "loss": 0.3298,
+      "step": 22104
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7613942048833674e-06,
+      "loss": 0.306,
+      "step": 22105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7606210161303316e-06,
+      "loss": 0.2902,
+      "step": 22106
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7598479183043e-06,
+      "loss": 0.2905,
+      "step": 22107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7590749114149827e-06,
+      "loss": 0.2976,
+      "step": 22108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7583019954720926e-06,
+      "loss": 0.3458,
+      "step": 22109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7575291704853325e-06,
+      "loss": 0.3375,
+      "step": 22110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7567564364644108e-06,
+      "loss": 0.3109,
+      "step": 22111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.755983793419037e-06,
+      "loss": 0.2609,
+      "step": 22112
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7552112413589106e-06,
+      "loss": 0.3201,
+      "step": 22113
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7544387802937357e-06,
+      "loss": 0.3276,
+      "step": 22114
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7536664102332177e-06,
+      "loss": 0.321,
+      "step": 22115
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.752894131187054e-06,
+      "loss": 0.2976,
+      "step": 22116
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.752121943164948e-06,
+      "loss": 0.3438,
+      "step": 22117
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.751349846176594e-06,
+      "loss": 0.3125,
+      "step": 22118
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.750577840231694e-06,
+      "loss": 0.3097,
+      "step": 22119
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.74980592533994e-06,
+      "loss": 0.3211,
+      "step": 22120
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7490341015110334e-06,
+      "loss": 0.3164,
+      "step": 22121
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7482623687546605e-06,
+      "loss": 0.3161,
+      "step": 22122
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7474907270805184e-06,
+      "loss": 0.3003,
+      "step": 22123
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7467191764983027e-06,
+      "loss": 0.3148,
+      "step": 22124
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7459477170176962e-06,
+      "loss": 0.2855,
+      "step": 22125
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.745176348648393e-06,
+      "loss": 0.3449,
+      "step": 22126
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744405071400085e-06,
+      "loss": 0.3173,
+      "step": 22127
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.743633885282452e-06,
+      "loss": 0.3043,
+      "step": 22128
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7428627903051863e-06,
+      "loss": 0.3233,
+      "step": 22129
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7420917864779684e-06,
+      "loss": 0.3003,
+      "step": 22130
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.741320873810487e-06,
+      "loss": 0.3159,
+      "step": 22131
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7405500523124185e-06,
+      "loss": 0.3306,
+      "step": 22132
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7397793219934475e-06,
+      "loss": 0.3312,
+      "step": 22133
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.739008682863259e-06,
+      "loss": 0.3293,
+      "step": 22134
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7382381349315244e-06,
+      "loss": 0.3226,
+      "step": 22135
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7374676782079257e-06,
+      "loss": 0.3235,
+      "step": 22136
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7366973127021424e-06,
+      "loss": 0.2991,
+      "step": 22137
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7359270384238455e-06,
+      "loss": 0.305,
+      "step": 22138
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7351568553827144e-06,
+      "loss": 0.3093,
+      "step": 22139
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.734386763588419e-06,
+      "loss": 0.2746,
+      "step": 22140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7336167630506317e-06,
+      "loss": 0.3114,
+      "step": 22141
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7328468537790233e-06,
+      "loss": 0.2725,
+      "step": 22142
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7320770357832695e-06,
+      "loss": 0.2743,
+      "step": 22143
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7313073090730313e-06,
+      "loss": 0.3121,
+      "step": 22144
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7305376736579804e-06,
+      "loss": 0.3107,
+      "step": 22145
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7297681295477872e-06,
+      "loss": 0.3228,
+      "step": 22146
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7289986767521105e-06,
+      "loss": 0.2945,
+      "step": 22147
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7282293152806162e-06,
+      "loss": 0.3029,
+      "step": 22148
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.727460045142973e-06,
+      "loss": 0.3445,
+      "step": 22149
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.726690866348839e-06,
+      "loss": 0.3242,
+      "step": 22150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725921778907872e-06,
+      "loss": 0.3246,
+      "step": 22151
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7251527828297376e-06,
+      "loss": 0.3164,
+      "step": 22152
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7243838781240894e-06,
+      "loss": 0.2957,
+      "step": 22153
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7236150648005864e-06,
+      "loss": 0.2995,
+      "step": 22154
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.722846342868889e-06,
+      "loss": 0.3147,
+      "step": 22155
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7220777123386465e-06,
+      "loss": 0.3075,
+      "step": 22156
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.721309173219515e-06,
+      "loss": 0.308,
+      "step": 22157
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7205407255211513e-06,
+      "loss": 0.3118,
+      "step": 22158
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7197723692531996e-06,
+      "loss": 0.3006,
+      "step": 22159
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.719004104425319e-06,
+      "loss": 0.3031,
+      "step": 22160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.718235931047154e-06,
+      "loss": 0.3287,
+      "step": 22161
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7174678491283513e-06,
+      "loss": 0.3337,
+      "step": 22162
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7166998586785597e-06,
+      "loss": 0.32,
+      "step": 22163
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7159319597074285e-06,
+      "loss": 0.3073,
+      "step": 22164
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7151641522245974e-06,
+      "loss": 0.2925,
+      "step": 22165
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7143964362397124e-06,
+      "loss": 0.3099,
+      "step": 22166
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7136288117624166e-06,
+      "loss": 0.3044,
+      "step": 22167
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7128612788023535e-06,
+      "loss": 0.3167,
+      "step": 22168
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7120938373691575e-06,
+      "loss": 0.3244,
+      "step": 22169
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7113264874724744e-06,
+      "loss": 0.3108,
+      "step": 22170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7105592291219394e-06,
+      "loss": 0.3312,
+      "step": 22171
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7097920623271844e-06,
+      "loss": 0.3109,
+      "step": 22172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.70902498709785e-06,
+      "loss": 0.2938,
+      "step": 22173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7082580034435744e-06,
+      "loss": 0.3071,
+      "step": 22174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.707491111373982e-06,
+      "loss": 0.3267,
+      "step": 22175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7067243108987106e-06,
+      "loss": 0.3257,
+      "step": 22176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7059576020273924e-06,
+      "loss": 0.3085,
+      "step": 22177
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7051909847696534e-06,
+      "loss": 0.3208,
+      "step": 22178
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704424459135123e-06,
+      "loss": 0.3196,
+      "step": 22179
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7036580251334345e-06,
+      "loss": 0.3157,
+      "step": 22180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702891682774209e-06,
+      "loss": 0.3009,
+      "step": 22181
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7021254320670688e-06,
+      "loss": 0.3157,
+      "step": 22182
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7013592730216464e-06,
+      "loss": 0.293,
+      "step": 22183
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.700593205647556e-06,
+      "loss": 0.3289,
+      "step": 22184
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.699827229954425e-06,
+      "loss": 0.3023,
+      "step": 22185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6990613459518744e-06,
+      "loss": 0.3665,
+      "step": 22186
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6982955536495205e-06,
+      "loss": 0.3032,
+      "step": 22187
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6975298530569826e-06,
+      "loss": 0.3055,
+      "step": 22188
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6967642441838824e-06,
+      "loss": 0.2933,
+      "step": 22189
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.695998727039829e-06,
+      "loss": 0.2912,
+      "step": 22190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6952333016344434e-06,
+      "loss": 0.3474,
+      "step": 22191
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.694467967977337e-06,
+      "loss": 0.2965,
+      "step": 22192
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.693702726078119e-06,
+      "loss": 0.3008,
+      "step": 22193
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.692937575946404e-06,
+      "loss": 0.3013,
+      "step": 22194
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6921725175918045e-06,
+      "loss": 0.3185,
+      "step": 22195
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6914075510239246e-06,
+      "loss": 0.3121,
+      "step": 22196
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6906426762523764e-06,
+      "loss": 0.2922,
+      "step": 22197
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6898778932867677e-06,
+      "loss": 0.3126,
+      "step": 22198
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6891132021366993e-06,
+      "loss": 0.307,
+      "step": 22199
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688348602811778e-06,
+      "loss": 0.2801,
+      "step": 22200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6875840953216115e-06,
+      "loss": 0.3326,
+      "step": 22201
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6868196796757982e-06,
+      "loss": 0.3198,
+      "step": 22202
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6860553558839364e-06,
+      "loss": 0.3054,
+      "step": 22203
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6852911239556277e-06,
+      "loss": 0.3112,
+      "step": 22204
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.684526983900476e-06,
+      "loss": 0.3455,
+      "step": 22205
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6837629357280726e-06,
+      "loss": 0.343,
+      "step": 22206
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.682998979448016e-06,
+      "loss": 0.3263,
+      "step": 22207
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6822351150699056e-06,
+      "loss": 0.3065,
+      "step": 22208
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6814713426033277e-06,
+      "loss": 0.2877,
+      "step": 22209
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.680707662057881e-06,
+      "loss": 0.3008,
+      "step": 22210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.679944073443158e-06,
+      "loss": 0.3211,
+      "step": 22211
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6791805767687484e-06,
+      "loss": 0.2943,
+      "step": 22212
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6784171720442377e-06,
+      "loss": 0.3405,
+      "step": 22213
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.677653859279219e-06,
+      "loss": 0.3054,
+      "step": 22214
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.676890638483276e-06,
+      "loss": 0.3236,
+      "step": 22215
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.676127509665997e-06,
+      "loss": 0.2945,
+      "step": 22216
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6753644728369686e-06,
+      "loss": 0.3251,
+      "step": 22217
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6746015280057702e-06,
+      "loss": 0.3113,
+      "step": 22218
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6738386751819866e-06,
+      "loss": 0.3165,
+      "step": 22219
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6730759143752038e-06,
+      "loss": 0.318,
+      "step": 22220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.672313245594993e-06,
+      "loss": 0.3438,
+      "step": 22221
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.671550668850943e-06,
+      "loss": 0.3492,
+      "step": 22222
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.670788184152623e-06,
+      "loss": 0.3267,
+      "step": 22223
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6700257915096173e-06,
+      "loss": 0.3068,
+      "step": 22224
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669263490931494e-06,
+      "loss": 0.2693,
+      "step": 22225
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6685012824278365e-06,
+      "loss": 0.3178,
+      "step": 22226
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6677391660082097e-06,
+      "loss": 0.3256,
+      "step": 22227
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6669771416821898e-06,
+      "loss": 0.3185,
+      "step": 22228
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6662152094593506e-06,
+      "loss": 0.3009,
+      "step": 22229
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6654533693492557e-06,
+      "loss": 0.283,
+      "step": 22230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.664691621361478e-06,
+      "loss": 0.354,
+      "step": 22231
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6639299655055874e-06,
+      "loss": 0.2937,
+      "step": 22232
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.663168401791143e-06,
+      "loss": 0.3101,
+      "step": 22233
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662406930227719e-06,
+      "loss": 0.3269,
+      "step": 22234
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.661645550824874e-06,
+      "loss": 0.2702,
+      "step": 22235
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6608842635921696e-06,
+      "loss": 0.3029,
+      "step": 22236
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6601230685391686e-06,
+      "loss": 0.3267,
+      "step": 22237
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6593619656754345e-06,
+      "loss": 0.3048,
+      "step": 22238
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6586009550105274e-06,
+      "loss": 0.2941,
+      "step": 22239
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6578400365540004e-06,
+      "loss": 0.3257,
+      "step": 22240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.657079210315413e-06,
+      "loss": 0.347,
+      "step": 22241
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.656318476304326e-06,
+      "loss": 0.3327,
+      "step": 22242
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.655557834530288e-06,
+      "loss": 0.319,
+      "step": 22243
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.654797285002857e-06,
+      "loss": 0.2883,
+      "step": 22244
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.654036827731583e-06,
+      "loss": 0.3046,
+      "step": 22245
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6532764627260153e-06,
+      "loss": 0.3457,
+      "step": 22246
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6525161899957065e-06,
+      "loss": 0.3312,
+      "step": 22247
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6517560095502095e-06,
+      "loss": 0.2899,
+      "step": 22248
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6509959213990644e-06,
+      "loss": 0.3625,
+      "step": 22249
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6502359255518227e-06,
+      "loss": 0.2965,
+      "step": 22250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6494760220180336e-06,
+      "loss": 0.3068,
+      "step": 22251
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.648716210807233e-06,
+      "loss": 0.3142,
+      "step": 22252
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.647956491928969e-06,
+      "loss": 0.3014,
+      "step": 22253
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.647196865392786e-06,
+      "loss": 0.3151,
+      "step": 22254
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6464373312082225e-06,
+      "loss": 0.3352,
+      "step": 22255
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6456778893848144e-06,
+      "loss": 0.3206,
+      "step": 22256
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6449185399321076e-06,
+      "loss": 0.3215,
+      "step": 22257
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.644159282859632e-06,
+      "loss": 0.2884,
+      "step": 22258
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.643400118176929e-06,
+      "loss": 0.308,
+      "step": 22259
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642641045893534e-06,
+      "loss": 0.2832,
+      "step": 22260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641882066018977e-06,
+      "loss": 0.3046,
+      "step": 22261
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6411231785627945e-06,
+      "loss": 0.3146,
+      "step": 22262
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6403643835345183e-06,
+      "loss": 0.3209,
+      "step": 22263
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6396056809436756e-06,
+      "loss": 0.3509,
+      "step": 22264
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6388470707998006e-06,
+      "loss": 0.3317,
+      "step": 22265
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.638088553112419e-06,
+      "loss": 0.3266,
+      "step": 22266
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6373301278910548e-06,
+      "loss": 0.3407,
+      "step": 22267
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6365717951452354e-06,
+      "loss": 0.287,
+      "step": 22268
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6358135548844897e-06,
+      "loss": 0.274,
+      "step": 22269
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6350554071183365e-06,
+      "loss": 0.3008,
+      "step": 22270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.634297351856299e-06,
+      "loss": 0.3194,
+      "step": 22271
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633539389107902e-06,
+      "loss": 0.2806,
+      "step": 22272
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.63278151888266e-06,
+      "loss": 0.3328,
+      "step": 22273
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.632023741190095e-06,
+      "loss": 0.3058,
+      "step": 22274
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6312660560397273e-06,
+      "loss": 0.3163,
+      "step": 22275
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6305084634410716e-06,
+      "loss": 0.3147,
+      "step": 22276
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6297509634036378e-06,
+      "loss": 0.3074,
+      "step": 22277
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6289935559369452e-06,
+      "loss": 0.3387,
+      "step": 22278
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6282362410505104e-06,
+      "loss": 0.2715,
+      "step": 22279
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6274790187538366e-06,
+      "loss": 0.2802,
+      "step": 22280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6267218890564406e-06,
+      "loss": 0.2832,
+      "step": 22281
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6259648519678326e-06,
+      "loss": 0.3043,
+      "step": 22282
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.625207907497517e-06,
+      "loss": 0.3214,
+      "step": 22283
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6244510556550017e-06,
+      "loss": 0.3247,
+      "step": 22284
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.623694296449798e-06,
+      "loss": 0.3071,
+      "step": 22285
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6229376298914077e-06,
+      "loss": 0.3052,
+      "step": 22286
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6221810559893303e-06,
+      "loss": 0.3408,
+      "step": 22287
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6214245747530754e-06,
+      "loss": 0.2977,
+      "step": 22288
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6206681861921377e-06,
+      "loss": 0.2984,
+      "step": 22289
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6199118903160215e-06,
+      "loss": 0.2815,
+      "step": 22290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6191556871342273e-06,
+      "loss": 0.346,
+      "step": 22291
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6183995766562485e-06,
+      "loss": 0.344,
+      "step": 22292
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.617643558891585e-06,
+      "loss": 0.3281,
+      "step": 22293
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.616887633849734e-06,
+      "loss": 0.2768,
+      "step": 22294
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.616131801540185e-06,
+      "loss": 0.2861,
+      "step": 22295
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6153760619724377e-06,
+      "loss": 0.3265,
+      "step": 22296
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6146204151559794e-06,
+      "loss": 0.3195,
+      "step": 22297
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6138648611003005e-06,
+      "loss": 0.2814,
+      "step": 22298
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6131093998148926e-06,
+      "loss": 0.3015,
+      "step": 22299
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6123540313092466e-06,
+      "loss": 0.3096,
+      "step": 22300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6115987555928447e-06,
+      "loss": 0.31,
+      "step": 22301
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.610843572675177e-06,
+      "loss": 0.328,
+      "step": 22302
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.61008848256573e-06,
+      "loss": 0.3047,
+      "step": 22303
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6093334852739838e-06,
+      "loss": 0.3044,
+      "step": 22304
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6085785808094223e-06,
+      "loss": 0.3236,
+      "step": 22305
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.60782376918153e-06,
+      "loss": 0.3049,
+      "step": 22306
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6070690503997866e-06,
+      "loss": 0.3177,
+      "step": 22307
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.606314424473666e-06,
+      "loss": 0.3265,
+      "step": 22308
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6055598914126532e-06,
+      "loss": 0.3303,
+      "step": 22309
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.60480545122622e-06,
+      "loss": 0.2963,
+      "step": 22310
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6040511039238435e-06,
+      "loss": 0.3359,
+      "step": 22311
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6032968495150003e-06,
+      "loss": 0.3156,
+      "step": 22312
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6025426880091665e-06,
+      "loss": 0.3062,
+      "step": 22313
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.601788619415806e-06,
+      "loss": 0.3169,
+      "step": 22314
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6010346437443956e-06,
+      "loss": 0.3108,
+      "step": 22315
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.600280761004408e-06,
+      "loss": 0.3188,
+      "step": 22316
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5995269712053085e-06,
+      "loss": 0.2934,
+      "step": 22317
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.598773274356561e-06,
+      "loss": 0.269,
+      "step": 22318
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5980196704676388e-06,
+      "loss": 0.2887,
+      "step": 22319
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.597266159548002e-06,
+      "loss": 0.3099,
+      "step": 22320
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5965127416071155e-06,
+      "loss": 0.341,
+      "step": 22321
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5957594166544477e-06,
+      "loss": 0.3095,
+      "step": 22322
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5950061846994524e-06,
+      "loss": 0.281,
+      "step": 22323
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5942530457515945e-06,
+      "loss": 0.3406,
+      "step": 22324
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.593499999820337e-06,
+      "loss": 0.2545,
+      "step": 22325
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5927470469151305e-06,
+      "loss": 0.3075,
+      "step": 22326
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591994187045439e-06,
+      "loss": 0.3437,
+      "step": 22327
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5912414202207116e-06,
+      "loss": 0.3527,
+      "step": 22328
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5904887464504115e-06,
+      "loss": 0.3172,
+      "step": 22329
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5897361657439835e-06,
+      "loss": 0.3478,
+      "step": 22330
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.588983678110888e-06,
+      "loss": 0.296,
+      "step": 22331
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5882312835605695e-06,
+      "loss": 0.337,
+      "step": 22332
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5874789821024804e-06,
+      "loss": 0.3207,
+      "step": 22333
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5867267737460744e-06,
+      "loss": 0.3046,
+      "step": 22334
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.585974658500792e-06,
+      "loss": 0.3191,
+      "step": 22335
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5852226363760824e-06,
+      "loss": 0.3265,
+      "step": 22336
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5844707073813958e-06,
+      "loss": 0.2945,
+      "step": 22337
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5837188715261687e-06,
+      "loss": 0.2916,
+      "step": 22338
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.582967128819851e-06,
+      "loss": 0.2956,
+      "step": 22339
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5822154792718823e-06,
+      "loss": 0.3199,
+      "step": 22340
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5814639228916982e-06,
+      "loss": 0.2938,
+      "step": 22341
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5807124596887446e-06,
+      "loss": 0.3057,
+      "step": 22342
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.579961089672459e-06,
+      "loss": 0.3039,
+      "step": 22343
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.579209812852276e-06,
+      "loss": 0.3331,
+      "step": 22344
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5784586292376335e-06,
+      "loss": 0.314,
+      "step": 22345
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.577707538837969e-06,
+      "loss": 0.3286,
+      "step": 22346
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5769565416627095e-06,
+      "loss": 0.3193,
+      "step": 22347
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576205637721293e-06,
+      "loss": 0.3391,
+      "step": 22348
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.575454827023152e-06,
+      "loss": 0.3255,
+      "step": 22349
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.574704109577714e-06,
+      "loss": 0.3468,
+      "step": 22350
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5739534853944048e-06,
+      "loss": 0.3002,
+      "step": 22351
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5732029544826564e-06,
+      "loss": 0.3112,
+      "step": 22352
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5724525168518975e-06,
+      "loss": 0.3045,
+      "step": 22353
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5717021725115487e-06,
+      "loss": 0.3058,
+      "step": 22354
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5709519214710367e-06,
+      "loss": 0.3033,
+      "step": 22355
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.570201763739787e-06,
+      "loss": 0.3467,
+      "step": 22356
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5694516993272155e-06,
+      "loss": 0.3136,
+      "step": 22357
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5687017282427486e-06,
+      "loss": 0.3047,
+      "step": 22358
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5679518504958055e-06,
+      "loss": 0.3034,
+      "step": 22359
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5672020660958043e-06,
+      "loss": 0.3114,
+      "step": 22360
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5664523750521586e-06,
+      "loss": 0.3014,
+      "step": 22361
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5657027773742904e-06,
+      "loss": 0.2857,
+      "step": 22362
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5649532730716075e-06,
+      "loss": 0.3105,
+      "step": 22363
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564203862153529e-06,
+      "loss": 0.2777,
+      "step": 22364
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.563454544629469e-06,
+      "loss": 0.3369,
+      "step": 22365
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5627053205088337e-06,
+      "loss": 0.3314,
+      "step": 22366
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.561956189801036e-06,
+      "loss": 0.3279,
+      "step": 22367
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.561207152515488e-06,
+      "loss": 0.2966,
+      "step": 22368
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5604582086615928e-06,
+      "loss": 0.3338,
+      "step": 22369
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.559709358248761e-06,
+      "loss": 0.3266,
+      "step": 22370
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5589606012863968e-06,
+      "loss": 0.2844,
+      "step": 22371
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.558211937783901e-06,
+      "loss": 0.3135,
+      "step": 22372
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.55746336775068e-06,
+      "loss": 0.3138,
+      "step": 22373
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5567148911961393e-06,
+      "loss": 0.3037,
+      "step": 22374
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5559665081296725e-06,
+      "loss": 0.3224,
+      "step": 22375
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5552182185606845e-06,
+      "loss": 0.3171,
+      "step": 22376
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.554470022498575e-06,
+      "loss": 0.3134,
+      "step": 22377
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5537219199527365e-06,
+      "loss": 0.2977,
+      "step": 22378
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5529739109325667e-06,
+      "loss": 0.306,
+      "step": 22379
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.552225995447465e-06,
+      "loss": 0.3314,
+      "step": 22380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551478173506823e-06,
+      "loss": 0.3059,
+      "step": 22381
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5507304451200277e-06,
+      "loss": 0.3387,
+      "step": 22382
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5499828102964796e-06,
+      "loss": 0.3134,
+      "step": 22383
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5492352690455602e-06,
+      "loss": 0.348,
+      "step": 22384
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548487821376664e-06,
+      "loss": 0.3271,
+      "step": 22385
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.547740467299179e-06,
+      "loss": 0.3156,
+      "step": 22386
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5469932068224923e-06,
+      "loss": 0.2991,
+      "step": 22387
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5462460399559865e-06,
+      "loss": 0.3123,
+      "step": 22388
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5454989667090478e-06,
+      "loss": 0.3171,
+      "step": 22389
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.544751987091062e-06,
+      "loss": 0.3018,
+      "step": 22390
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5440051011114097e-06,
+      "loss": 0.3089,
+      "step": 22391
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.543258308779467e-06,
+      "loss": 0.3105,
+      "step": 22392
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.542511610104621e-06,
+      "loss": 0.325,
+      "step": 22393
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.541765005096244e-06,
+      "loss": 0.3093,
+      "step": 22394
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.541018493763716e-06,
+      "loss": 0.3235,
+      "step": 22395
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5402720761164166e-06,
+      "loss": 0.3212,
+      "step": 22396
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.539525752163714e-06,
+      "loss": 0.3082,
+      "step": 22397
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.538779521914987e-06,
+      "loss": 0.3573,
+      "step": 22398
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5380333853796092e-06,
+      "loss": 0.3207,
+      "step": 22399
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5372873425669474e-06,
+      "loss": 0.3194,
+      "step": 22400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.536541393486377e-06,
+      "loss": 0.3098,
+      "step": 22401
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5357955381472622e-06,
+      "loss": 0.3316,
+      "step": 22402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5350497765589765e-06,
+      "loss": 0.2966,
+      "step": 22403
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5343041087308807e-06,
+      "loss": 0.3117,
+      "step": 22404
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.533558534672346e-06,
+      "loss": 0.3227,
+      "step": 22405
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.532813054392732e-06,
+      "loss": 0.3329,
+      "step": 22406
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.532067667901403e-06,
+      "loss": 0.3267,
+      "step": 22407
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5313223752077265e-06,
+      "loss": 0.2848,
+      "step": 22408
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.530577176321056e-06,
+      "loss": 0.3122,
+      "step": 22409
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.529832071250754e-06,
+      "loss": 0.3075,
+      "step": 22410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5290870600061835e-06,
+      "loss": 0.3013,
+      "step": 22411
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5283421425966947e-06,
+      "loss": 0.3143,
+      "step": 22412
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5275973190316492e-06,
+      "loss": 0.3063,
+      "step": 22413
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.526852589320401e-06,
+      "loss": 0.3099,
+      "step": 22414
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5261079534722986e-06,
+      "loss": 0.2586,
+      "step": 22415
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5253634114966996e-06,
+      "loss": 0.3246,
+      "step": 22416
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5246189634029584e-06,
+      "loss": 0.3129,
+      "step": 22417
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5238746092004175e-06,
+      "loss": 0.3018,
+      "step": 22418
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5231303488984305e-06,
+      "loss": 0.3289,
+      "step": 22419
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.522386182506349e-06,
+      "loss": 0.3005,
+      "step": 22420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.521642110033512e-06,
+      "loss": 0.3242,
+      "step": 22421
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.520898131489269e-06,
+      "loss": 0.3449,
+      "step": 22422
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5201542468829678e-06,
+      "loss": 0.3075,
+      "step": 22423
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5194104562239472e-06,
+      "loss": 0.3065,
+      "step": 22424
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5186667595215473e-06,
+      "loss": 0.3178,
+      "step": 22425
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5179231567851125e-06,
+      "loss": 0.3054,
+      "step": 22426
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.517179648023985e-06,
+      "loss": 0.3065,
+      "step": 22427
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5164362332474977e-06,
+      "loss": 0.31,
+      "step": 22428
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5156929124649898e-06,
+      "loss": 0.3391,
+      "step": 22429
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5149496856858e-06,
+      "loss": 0.313,
+      "step": 22430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5142065529192603e-06,
+      "loss": 0.3187,
+      "step": 22431
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5134635141747045e-06,
+      "loss": 0.2954,
+      "step": 22432
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.51272056946147e-06,
+      "loss": 0.3243,
+      "step": 22433
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5119777187888826e-06,
+      "loss": 0.3169,
+      "step": 22434
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5112349621662736e-06,
+      "loss": 0.2919,
+      "step": 22435
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5104922996029744e-06,
+      "loss": 0.3019,
+      "step": 22436
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5097497311083096e-06,
+      "loss": 0.3258,
+      "step": 22437
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5090072566916067e-06,
+      "loss": 0.3192,
+      "step": 22438
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5082648763621954e-06,
+      "loss": 0.2966,
+      "step": 22439
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507522590129394e-06,
+      "loss": 0.3094,
+      "step": 22440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5067803980025285e-06,
+      "loss": 0.3321,
+      "step": 22441
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.506038299990924e-06,
+      "loss": 0.2953,
+      "step": 22442
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.505296296103895e-06,
+      "loss": 0.3194,
+      "step": 22443
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.504554386350767e-06,
+      "loss": 0.3253,
+      "step": 22444
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5038125707408567e-06,
+      "loss": 0.3159,
+      "step": 22445
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5030708492834763e-06,
+      "loss": 0.3101,
+      "step": 22446
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502329221987947e-06,
+      "loss": 0.3236,
+      "step": 22447
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5015876888635837e-06,
+      "loss": 0.2714,
+      "step": 22448
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.500846249919697e-06,
+      "loss": 0.2845,
+      "step": 22449
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5001049051656025e-06,
+      "loss": 0.3407,
+      "step": 22450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4993636546106127e-06,
+      "loss": 0.3068,
+      "step": 22451
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498622498264032e-06,
+      "loss": 0.3426,
+      "step": 22452
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.497881436135173e-06,
+      "loss": 0.3154,
+      "step": 22453
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4971404682333466e-06,
+      "loss": 0.3188,
+      "step": 22454
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4963995945678555e-06,
+      "loss": 0.3176,
+      "step": 22455
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4956588151480032e-06,
+      "loss": 0.3116,
+      "step": 22456
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494918129983096e-06,
+      "loss": 0.3076,
+      "step": 22457
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494177539082442e-06,
+      "loss": 0.3407,
+      "step": 22458
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.493437042455333e-06,
+      "loss": 0.3467,
+      "step": 22459
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4926966401110773e-06,
+      "loss": 0.3405,
+      "step": 22460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4919563320589747e-06,
+      "loss": 0.3342,
+      "step": 22461
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4912161183083172e-06,
+      "loss": 0.3171,
+      "step": 22462
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4904759988684066e-06,
+      "loss": 0.3124,
+      "step": 22463
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4897359737485404e-06,
+      "loss": 0.2822,
+      "step": 22464
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4889960429580117e-06,
+      "loss": 0.2884,
+      "step": 22465
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.48825620650611e-06,
+      "loss": 0.3242,
+      "step": 22466
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4875164644021343e-06,
+      "loss": 0.3021,
+      "step": 22467
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4867768166553695e-06,
+      "loss": 0.3202,
+      "step": 22468
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4860372632751084e-06,
+      "loss": 0.3058,
+      "step": 22469
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4852978042706434e-06,
+      "loss": 0.3046,
+      "step": 22470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4845584396512557e-06,
+      "loss": 0.338,
+      "step": 22471
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.483819169426236e-06,
+      "loss": 0.2759,
+      "step": 22472
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4830799936048698e-06,
+      "loss": 0.309,
+      "step": 22473
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4823409121964372e-06,
+      "loss": 0.299,
+      "step": 22474
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4816019252102274e-06,
+      "loss": 0.3118,
+      "step": 22475
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4808630326555184e-06,
+      "loss": 0.3073,
+      "step": 22476
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4801242345415876e-06,
+      "loss": 0.3192,
+      "step": 22477
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.479385530877717e-06,
+      "loss": 0.2879,
+      "step": 22478
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478646921673189e-06,
+      "loss": 0.3296,
+      "step": 22479
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.477908406937274e-06,
+      "loss": 0.3361,
+      "step": 22480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.477169986679251e-06,
+      "loss": 0.3138,
+      "step": 22481
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.476431660908396e-06,
+      "loss": 0.3272,
+      "step": 22482
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4756934296339784e-06,
+      "loss": 0.317,
+      "step": 22483
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474955292865273e-06,
+      "loss": 0.2921,
+      "step": 22484
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4742172506115536e-06,
+      "loss": 0.2993,
+      "step": 22485
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4734793028820868e-06,
+      "loss": 0.2876,
+      "step": 22486
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.472741449686139e-06,
+      "loss": 0.296,
+      "step": 22487
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4720036910329824e-06,
+      "loss": 0.3217,
+      "step": 22488
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4712660269318787e-06,
+      "loss": 0.327,
+      "step": 22489
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.470528457392095e-06,
+      "loss": 0.2965,
+      "step": 22490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4697909824228983e-06,
+      "loss": 0.3304,
+      "step": 22491
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4690536020335466e-06,
+      "loss": 0.2902,
+      "step": 22492
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4683163162333026e-06,
+      "loss": 0.3346,
+      "step": 22493
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4675791250314273e-06,
+      "loss": 0.2798,
+      "step": 22494
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.466842028437183e-06,
+      "loss": 0.3214,
+      "step": 22495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.466105026459825e-06,
+      "loss": 0.3373,
+      "step": 22496
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4653681191086076e-06,
+      "loss": 0.3072,
+      "step": 22497
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4646313063927907e-06,
+      "loss": 0.345,
+      "step": 22498
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4638945883216236e-06,
+      "loss": 0.3442,
+      "step": 22499
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4631579649043626e-06,
+      "loss": 0.3017,
+      "step": 22500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.462421436150263e-06,
+      "loss": 0.2871,
+      "step": 22501
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461685002068569e-06,
+      "loss": 0.3358,
+      "step": 22502
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.460948662668534e-06,
+      "loss": 0.3002,
+      "step": 22503
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.460212417959409e-06,
+      "loss": 0.3045,
+      "step": 22504
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.459476267950435e-06,
+      "loss": 0.2986,
+      "step": 22505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458740212650864e-06,
+      "loss": 0.3299,
+      "step": 22506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4580042520699355e-06,
+      "loss": 0.2988,
+      "step": 22507
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4572683862169e-06,
+      "loss": 0.3414,
+      "step": 22508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4565326151009917e-06,
+      "loss": 0.3259,
+      "step": 22509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4557969387314595e-06,
+      "loss": 0.3193,
+      "step": 22510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4550613571175384e-06,
+      "loss": 0.2992,
+      "step": 22511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454325870268468e-06,
+      "loss": 0.3212,
+      "step": 22512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4535904781934904e-06,
+      "loss": 0.3503,
+      "step": 22513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4528551809018363e-06,
+      "loss": 0.2899,
+      "step": 22514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.452119978402744e-06,
+      "loss": 0.3308,
+      "step": 22515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.45138487070545e-06,
+      "loss": 0.3061,
+      "step": 22516
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4506498578191827e-06,
+      "loss": 0.2715,
+      "step": 22517
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4499149397531784e-06,
+      "loss": 0.2877,
+      "step": 22518
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.449180116516665e-06,
+      "loss": 0.2857,
+      "step": 22519
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4484453881188697e-06,
+      "loss": 0.2878,
+      "step": 22520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4477107545690225e-06,
+      "loss": 0.3421,
+      "step": 22521
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4469762158763554e-06,
+      "loss": 0.3136,
+      "step": 22522
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.446241772050086e-06,
+      "loss": 0.3305,
+      "step": 22523
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4455074230994434e-06,
+      "loss": 0.3104,
+      "step": 22524
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.444773169033654e-06,
+      "loss": 0.3283,
+      "step": 22525
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4440390098619337e-06,
+      "loss": 0.3061,
+      "step": 22526
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4433049455935067e-06,
+      "loss": 0.2889,
+      "step": 22527
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4425709762375947e-06,
+      "loss": 0.3066,
+      "step": 22528
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.441837101803415e-06,
+      "loss": 0.3158,
+      "step": 22529
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4411033223001823e-06,
+      "loss": 0.3314,
+      "step": 22530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4403696377371146e-06,
+      "loss": 0.3261,
+      "step": 22531
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4396360481234303e-06,
+      "loss": 0.307,
+      "step": 22532
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4389025534683384e-06,
+      "loss": 0.3233,
+      "step": 22533
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4381691537810538e-06,
+      "loss": 0.3233,
+      "step": 22534
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4374358490707917e-06,
+      "loss": 0.3173,
+      "step": 22535
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4367026393467552e-06,
+      "loss": 0.2893,
+      "step": 22536
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.435969524618157e-06,
+      "loss": 0.3064,
+      "step": 22537
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4352365048942083e-06,
+      "loss": 0.2906,
+      "step": 22538
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434503580184112e-06,
+      "loss": 0.3,
+      "step": 22539
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.433770750497073e-06,
+      "loss": 0.2993,
+      "step": 22540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4330380158422995e-06,
+      "loss": 0.314,
+      "step": 22541
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4323053762289894e-06,
+      "loss": 0.2765,
+      "step": 22542
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.431572831666347e-06,
+      "loss": 0.3442,
+      "step": 22543
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.430840382163577e-06,
+      "loss": 0.327,
+      "step": 22544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.430108027729873e-06,
+      "loss": 0.3441,
+      "step": 22545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4293757683744357e-06,
+      "loss": 0.3001,
+      "step": 22546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4286436041064667e-06,
+      "loss": 0.3327,
+      "step": 22547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4279115349351546e-06,
+      "loss": 0.3135,
+      "step": 22548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4271795608697e-06,
+      "loss": 0.3133,
+      "step": 22549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.426447681919295e-06,
+      "loss": 0.3215,
+      "step": 22550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4257158980931272e-06,
+      "loss": 0.2988,
+      "step": 22551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4249842094003928e-06,
+      "loss": 0.3055,
+      "step": 22552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4242526158502843e-06,
+      "loss": 0.3169,
+      "step": 22553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4235211174519844e-06,
+      "loss": 0.3003,
+      "step": 22554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4227897142146827e-06,
+      "loss": 0.3175,
+      "step": 22555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4220584061475707e-06,
+      "loss": 0.3004,
+      "step": 22556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4213271932598247e-06,
+      "loss": 0.2877,
+      "step": 22557
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.420596075560635e-06,
+      "loss": 0.2954,
+      "step": 22558
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.419865053059186e-06,
+      "loss": 0.3241,
+      "step": 22559
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4191341257646563e-06,
+      "loss": 0.3224,
+      "step": 22560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4184032936862233e-06,
+      "loss": 0.3121,
+      "step": 22561
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4176725568330727e-06,
+      "loss": 0.3317,
+      "step": 22562
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.416941915214377e-06,
+      "loss": 0.3124,
+      "step": 22563
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.416211368839314e-06,
+      "loss": 0.3341,
+      "step": 22564
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4154809177170656e-06,
+      "loss": 0.3355,
+      "step": 22565
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.414750561856798e-06,
+      "loss": 0.3218,
+      "step": 22566
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4140203012676878e-06,
+      "loss": 0.305,
+      "step": 22567
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.413290135958908e-06,
+      "loss": 0.3087,
+      "step": 22568
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.412560065939632e-06,
+      "loss": 0.2934,
+      "step": 22569
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.411830091219026e-06,
+      "loss": 0.2847,
+      "step": 22570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4111002118062565e-06,
+      "loss": 0.2822,
+      "step": 22571
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4103704277104965e-06,
+      "loss": 0.2842,
+      "step": 22572
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409640738940906e-06,
+      "loss": 0.3339,
+      "step": 22573
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4089111455066528e-06,
+      "loss": 0.3173,
+      "step": 22574
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.408181647416904e-06,
+      "loss": 0.344,
+      "step": 22575
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.407452244680815e-06,
+      "loss": 0.3289,
+      "step": 22576
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406722937307552e-06,
+      "loss": 0.3453,
+      "step": 22577
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.405993725306276e-06,
+      "loss": 0.3199,
+      "step": 22578
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.405264608686142e-06,
+      "loss": 0.3367,
+      "step": 22579
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4045355874563124e-06,
+      "loss": 0.3218,
+      "step": 22580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4038066616259426e-06,
+      "loss": 0.315,
+      "step": 22581
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4030778312041824e-06,
+      "loss": 0.3088,
+      "step": 22582
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4023490962001904e-06,
+      "loss": 0.2744,
+      "step": 22583
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4016204566231216e-06,
+      "loss": 0.307,
+      "step": 22584
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4008919124821227e-06,
+      "loss": 0.3366,
+      "step": 22585
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.400163463786348e-06,
+      "loss": 0.3404,
+      "step": 22586
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3994351105449475e-06,
+      "loss": 0.3471,
+      "step": 22587
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3987068527670653e-06,
+      "loss": 0.3255,
+      "step": 22588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.397978690461851e-06,
+      "loss": 0.3152,
+      "step": 22589
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.397250623638453e-06,
+      "loss": 0.2944,
+      "step": 22590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3965226523060125e-06,
+      "loss": 0.2931,
+      "step": 22591
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.395794776473671e-06,
+      "loss": 0.2945,
+      "step": 22592
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3950669961505744e-06,
+      "loss": 0.3287,
+      "step": 22593
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3943393113458614e-06,
+      "loss": 0.3365,
+      "step": 22594
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3936117220686717e-06,
+      "loss": 0.3457,
+      "step": 22595
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3928842283281473e-06,
+      "loss": 0.3027,
+      "step": 22596
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.392156830133421e-06,
+      "loss": 0.3133,
+      "step": 22597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3914295274936306e-06,
+      "loss": 0.2851,
+      "step": 22598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3907023204179135e-06,
+      "loss": 0.3351,
+      "step": 22599
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3899752089153994e-06,
+      "loss": 0.3059,
+      "step": 22600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.389248192995226e-06,
+      "loss": 0.3014,
+      "step": 22601
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3885212726665185e-06,
+      "loss": 0.3116,
+      "step": 22602
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387794447938413e-06,
+      "loss": 0.3059,
+      "step": 22603
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387067718820032e-06,
+      "loss": 0.316,
+      "step": 22604
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3863410853205073e-06,
+      "loss": 0.2857,
+      "step": 22605
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3856145474489677e-06,
+      "loss": 0.3361,
+      "step": 22606
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3848881052145333e-06,
+      "loss": 0.2652,
+      "step": 22607
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.384161758626331e-06,
+      "loss": 0.3049,
+      "step": 22608
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3834355076934854e-06,
+      "loss": 0.3425,
+      "step": 22609
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3827093524251145e-06,
+      "loss": 0.3261,
+      "step": 22610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.381983292830343e-06,
+      "loss": 0.3021,
+      "step": 22611
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.381257328918286e-06,
+      "loss": 0.3178,
+      "step": 22612
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380531460698066e-06,
+      "loss": 0.3341,
+      "step": 22613
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3798056881787946e-06,
+      "loss": 0.3015,
+      "step": 22614
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3790800113695933e-06,
+      "loss": 0.3109,
+      "step": 22615
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3783544302795713e-06,
+      "loss": 0.2859,
+      "step": 22616
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3776289449178448e-06,
+      "loss": 0.315,
+      "step": 22617
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.376903555293528e-06,
+      "loss": 0.2784,
+      "step": 22618
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3761782614157268e-06,
+      "loss": 0.3344,
+      "step": 22619
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3754530632935537e-06,
+      "loss": 0.3169,
+      "step": 22620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3747279609361197e-06,
+      "loss": 0.3401,
+      "step": 22621
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3740029543525266e-06,
+      "loss": 0.2949,
+      "step": 22622
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3732780435518863e-06,
+      "loss": 0.2858,
+      "step": 22623
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.372553228543302e-06,
+      "loss": 0.3019,
+      "step": 22624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371828509335873e-06,
+      "loss": 0.3128,
+      "step": 22625
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371103885938705e-06,
+      "loss": 0.3173,
+      "step": 22626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3703793583609013e-06,
+      "loss": 0.3221,
+      "step": 22627
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3696549266115588e-06,
+      "loss": 0.2757,
+      "step": 22628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3689305906997772e-06,
+      "loss": 0.3081,
+      "step": 22629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3682063506346576e-06,
+      "loss": 0.3206,
+      "step": 22630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3674822064252912e-06,
+      "loss": 0.291,
+      "step": 22631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366758158080775e-06,
+      "loss": 0.3048,
+      "step": 22632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366034205610207e-06,
+      "loss": 0.3211,
+      "step": 22633
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365310349022677e-06,
+      "loss": 0.2897,
+      "step": 22634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3645865883272735e-06,
+      "loss": 0.3,
+      "step": 22635
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3638629235330935e-06,
+      "loss": 0.2899,
+      "step": 22636
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.36313935464922e-06,
+      "loss": 0.3285,
+      "step": 22637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.362415881684743e-06,
+      "loss": 0.3141,
+      "step": 22638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3616925046487536e-06,
+      "loss": 0.302,
+      "step": 22639
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.360969223550331e-06,
+      "loss": 0.3074,
+      "step": 22640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3602460383985635e-06,
+      "loss": 0.2992,
+      "step": 22641
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.359522949202534e-06,
+      "loss": 0.3276,
+      "step": 22642
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358799955971326e-06,
+      "loss": 0.3164,
+      "step": 22643
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358077058714019e-06,
+      "loss": 0.3092,
+      "step": 22644
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357354257439689e-06,
+      "loss": 0.3149,
+      "step": 22645
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.356631552157421e-06,
+      "loss": 0.2843,
+      "step": 22646
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355908942876285e-06,
+      "loss": 0.2794,
+      "step": 22647
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355186429605362e-06,
+      "loss": 0.3029,
+      "step": 22648
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3544640123537277e-06,
+      "loss": 0.2945,
+      "step": 22649
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.35374169113045e-06,
+      "loss": 0.3453,
+      "step": 22650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.353019465944606e-06,
+      "loss": 0.2747,
+      "step": 22651
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3522973368052683e-06,
+      "loss": 0.352,
+      "step": 22652
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3515753037215018e-06,
+      "loss": 0.3408,
+      "step": 22653
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3508533667023803e-06,
+      "loss": 0.3272,
+      "step": 22654
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.350131525756969e-06,
+      "loss": 0.3057,
+      "step": 22655
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3494097808943315e-06,
+      "loss": 0.3168,
+      "step": 22656
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3486881321235357e-06,
+      "loss": 0.3067,
+      "step": 22657
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3479665794536478e-06,
+      "loss": 0.3206,
+      "step": 22658
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3472451228937254e-06,
+      "loss": 0.2954,
+      "step": 22659
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3465237624528335e-06,
+      "loss": 0.3218,
+      "step": 22660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345802498140034e-06,
+      "loss": 0.2974,
+      "step": 22661
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3450813299643815e-06,
+      "loss": 0.3007,
+      "step": 22662
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3443602579349356e-06,
+      "loss": 0.3037,
+      "step": 22663
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.343639282060758e-06,
+      "loss": 0.2597,
+      "step": 22664
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342918402350898e-06,
+      "loss": 0.3288,
+      "step": 22665
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3421976188144093e-06,
+      "loss": 0.3384,
+      "step": 22666
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.341476931460351e-06,
+      "loss": 0.3185,
+      "step": 22667
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3407563402977674e-06,
+      "loss": 0.2963,
+      "step": 22668
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3400358453357143e-06,
+      "loss": 0.3276,
+      "step": 22669
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3393154465832424e-06,
+      "loss": 0.3061,
+      "step": 22670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.338595144049395e-06,
+      "loss": 0.3213,
+      "step": 22671
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3378749377432207e-06,
+      "loss": 0.3049,
+      "step": 22672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.33715482767377e-06,
+      "loss": 0.3307,
+      "step": 22673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3364348138500803e-06,
+      "loss": 0.2844,
+      "step": 22674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3357148962812016e-06,
+      "loss": 0.3228,
+      "step": 22675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3349950749761705e-06,
+      "loss": 0.3252,
+      "step": 22676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.334275349944033e-06,
+      "loss": 0.3079,
+      "step": 22677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333555721193824e-06,
+      "loss": 0.3188,
+      "step": 22678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.332836188734584e-06,
+      "loss": 0.2974,
+      "step": 22679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.332116752575354e-06,
+      "loss": 0.3257,
+      "step": 22680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331397412725165e-06,
+      "loss": 0.324,
+      "step": 22681
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3306781691930535e-06,
+      "loss": 0.3122,
+      "step": 22682
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3299590219880563e-06,
+      "loss": 0.3318,
+      "step": 22683
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3292399711192003e-06,
+      "loss": 0.2996,
+      "step": 22684
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.328521016595523e-06,
+      "loss": 0.3222,
+      "step": 22685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.327802158426048e-06,
+      "loss": 0.3157,
+      "step": 22686
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3270833966198113e-06,
+      "loss": 0.2773,
+      "step": 22687
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.326364731185834e-06,
+      "loss": 0.2818,
+      "step": 22688
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3256461621331474e-06,
+      "loss": 0.2965,
+      "step": 22689
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.324927689470773e-06,
+      "loss": 0.3341,
+      "step": 22690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.324209313207736e-06,
+      "loss": 0.3343,
+      "step": 22691
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3234910333530627e-06,
+      "loss": 0.2856,
+      "step": 22692
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3227728499157698e-06,
+      "loss": 0.3076,
+      "step": 22693
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3220547629048796e-06,
+      "loss": 0.3286,
+      "step": 22694
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3213367723294146e-06,
+      "loss": 0.3062,
+      "step": 22695
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.320618878198386e-06,
+      "loss": 0.3414,
+      "step": 22696
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319901080520818e-06,
+      "loss": 0.2897,
+      "step": 22697
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319183379305722e-06,
+      "loss": 0.31,
+      "step": 22698
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318465774562111e-06,
+      "loss": 0.3042,
+      "step": 22699
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.317748266298999e-06,
+      "loss": 0.3227,
+      "step": 22700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3170308545254026e-06,
+      "loss": 0.335,
+      "step": 22701
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3163135392503264e-06,
+      "loss": 0.3489,
+      "step": 22702
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3155963204827814e-06,
+      "loss": 0.3302,
+      "step": 22703
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.314879198231781e-06,
+      "loss": 0.3109,
+      "step": 22704
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3141621725063248e-06,
+      "loss": 0.2779,
+      "step": 22705
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3134452433154254e-06,
+      "loss": 0.3284,
+      "step": 22706
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3127284106680814e-06,
+      "loss": 0.296,
+      "step": 22707
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3120116745733014e-06,
+      "loss": 0.3046,
+      "step": 22708
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311295035040083e-06,
+      "loss": 0.3245,
+      "step": 22709
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3105784920774322e-06,
+      "loss": 0.313,
+      "step": 22710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3098620456943434e-06,
+      "loss": 0.3099,
+      "step": 22711
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.309145695899817e-06,
+      "loss": 0.3138,
+      "step": 22712
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308429442702853e-06,
+      "loss": 0.3033,
+      "step": 22713
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3077132861124484e-06,
+      "loss": 0.3391,
+      "step": 22714
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306997226137593e-06,
+      "loss": 0.3212,
+      "step": 22715
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3062812627872856e-06,
+      "loss": 0.3157,
+      "step": 22716
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3055653960705136e-06,
+      "loss": 0.3024,
+      "step": 22717
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3048496259962737e-06,
+      "loss": 0.3049,
+      "step": 22718
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.304133952573552e-06,
+      "loss": 0.3015,
+      "step": 22719
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3034183758113404e-06,
+      "loss": 0.3428,
+      "step": 22720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3027028957186228e-06,
+      "loss": 0.2861,
+      "step": 22721
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.301987512304388e-06,
+      "loss": 0.3045,
+      "step": 22722
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3012722255776234e-06,
+      "loss": 0.3032,
+      "step": 22723
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3005570355473084e-06,
+      "loss": 0.2895,
+      "step": 22724
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.299841942222428e-06,
+      "loss": 0.3096,
+      "step": 22725
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.299126945611967e-06,
+      "loss": 0.3184,
+      "step": 22726
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2984120457249004e-06,
+      "loss": 0.3177,
+      "step": 22727
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297697242570213e-06,
+      "loss": 0.3354,
+      "step": 22728
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296982536156879e-06,
+      "loss": 0.331,
+      "step": 22729
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2962679264938726e-06,
+      "loss": 0.2915,
+      "step": 22730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2955534135901723e-06,
+      "loss": 0.2838,
+      "step": 22731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294838997454757e-06,
+      "loss": 0.344,
+      "step": 22732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2941246780965907e-06,
+      "loss": 0.3177,
+      "step": 22733
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2934104555246516e-06,
+      "loss": 0.3457,
+      "step": 22734
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292696329747911e-06,
+      "loss": 0.3169,
+      "step": 22735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291982300775333e-06,
+      "loss": 0.3084,
+      "step": 22736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.29126836861589e-06,
+      "loss": 0.3458,
+      "step": 22737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2905545332785496e-06,
+      "loss": 0.3327,
+      "step": 22738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.289840794772277e-06,
+      "loss": 0.3405,
+      "step": 22739
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2891271531060324e-06,
+      "loss": 0.3038,
+      "step": 22740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2884136082887855e-06,
+      "loss": 0.3106,
+      "step": 22741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2877001603294923e-06,
+      "loss": 0.3118,
+      "step": 22742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.286986809237118e-06,
+      "loss": 0.2707,
+      "step": 22743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.286273555020624e-06,
+      "loss": 0.3166,
+      "step": 22744
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.285560397688963e-06,
+      "loss": 0.3332,
+      "step": 22745
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284847337251096e-06,
+      "loss": 0.3251,
+      "step": 22746
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2841343737159804e-06,
+      "loss": 0.2724,
+      "step": 22747
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283421507092567e-06,
+      "loss": 0.3145,
+      "step": 22748
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2827087373898147e-06,
+      "loss": 0.3238,
+      "step": 22749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2819960646166693e-06,
+      "loss": 0.3375,
+      "step": 22750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281283488782089e-06,
+      "loss": 0.34,
+      "step": 22751
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2805710098950174e-06,
+      "loss": 0.3001,
+      "step": 22752
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2798586279644062e-06,
+      "loss": 0.3097,
+      "step": 22753
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2791463429992067e-06,
+      "loss": 0.3124,
+      "step": 22754
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2784341550083577e-06,
+      "loss": 0.3177,
+      "step": 22755
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.277722064000808e-06,
+      "loss": 0.3191,
+      "step": 22756
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.277010069985506e-06,
+      "loss": 0.2995,
+      "step": 22757
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2762981729713853e-06,
+      "loss": 0.3155,
+      "step": 22758
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2755863729673955e-06,
+      "loss": 0.3229,
+      "step": 22759
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2748746699824743e-06,
+      "loss": 0.3185,
+      "step": 22760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2741630640255563e-06,
+      "loss": 0.3182,
+      "step": 22761
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2734515551055836e-06,
+      "loss": 0.3186,
+      "step": 22762
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2727401432314944e-06,
+      "loss": 0.2725,
+      "step": 22763
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2720288284122196e-06,
+      "loss": 0.2869,
+      "step": 22764
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271317610656695e-06,
+      "loss": 0.2955,
+      "step": 22765
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2706064899738554e-06,
+      "loss": 0.3097,
+      "step": 22766
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.283,
+      "step": 22767
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2691845398619506e-06,
+      "loss": 0.3032,
+      "step": 22768
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.268473710450749e-06,
+      "loss": 0.338,
+      "step": 22769
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2677629781479503e-06,
+      "loss": 0.2898,
+      "step": 22770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2670523429624793e-06,
+      "loss": 0.3016,
+      "step": 22771
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.266341804903266e-06,
+      "loss": 0.3361,
+      "step": 22772
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2656313639792306e-06,
+      "loss": 0.3,
+      "step": 22773
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264921020199299e-06,
+      "loss": 0.2769,
+      "step": 22774
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264210773572394e-06,
+      "loss": 0.3199,
+      "step": 22775
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.263500624107433e-06,
+      "loss": 0.3315,
+      "step": 22776
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.262790571813337e-06,
+      "loss": 0.3017,
+      "step": 22777
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2620806166990282e-06,
+      "loss": 0.3292,
+      "step": 22778
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.261370758773417e-06,
+      "loss": 0.2961,
+      "step": 22779
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.260660998045425e-06,
+      "loss": 0.3066,
+      "step": 22780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.259951334523962e-06,
+      "loss": 0.324,
+      "step": 22781
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.259241768217947e-06,
+      "loss": 0.3117,
+      "step": 22782
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2585322991362847e-06,
+      "loss": 0.2966,
+      "step": 22783
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.257822927287894e-06,
+      "loss": 0.287,
+      "step": 22784
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2571136526816785e-06,
+      "loss": 0.3365,
+      "step": 22785
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.256404475326548e-06,
+      "loss": 0.3422,
+      "step": 22786
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.255695395231412e-06,
+      "loss": 0.3013,
+      "step": 22787
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.254986412405178e-06,
+      "loss": 0.3035,
+      "step": 22788
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.254277526856745e-06,
+      "loss": 0.3418,
+      "step": 22789
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2535687385950243e-06,
+      "loss": 0.3162,
+      "step": 22790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.252860047628911e-06,
+      "loss": 0.3301,
+      "step": 22791
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.252151453967312e-06,
+      "loss": 0.3125,
+      "step": 22792
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.251442957619121e-06,
+      "loss": 0.3538,
+      "step": 22793
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2507345585932438e-06,
+      "loss": 0.3194,
+      "step": 22794
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.250026256898572e-06,
+      "loss": 0.294,
+      "step": 22795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.249318052544004e-06,
+      "loss": 0.2946,
+      "step": 22796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.248609945538438e-06,
+      "loss": 0.3134,
+      "step": 22797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2479019358907628e-06,
+      "loss": 0.3288,
+      "step": 22798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2471940236098733e-06,
+      "loss": 0.3258,
+      "step": 22799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2464862087046644e-06,
+      "loss": 0.3164,
+      "step": 22800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2457784911840198e-06,
+      "loss": 0.3366,
+      "step": 22801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.245070871056835e-06,
+      "loss": 0.3009,
+      "step": 22802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2443633483319925e-06,
+      "loss": 0.2945,
+      "step": 22803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2436559230183794e-06,
+      "loss": 0.3066,
+      "step": 22804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242948595124882e-06,
+      "loss": 0.3347,
+      "step": 22805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242241364660388e-06,
+      "loss": 0.2913,
+      "step": 22806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.241534231633774e-06,
+      "loss": 0.31,
+      "step": 22807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240827196053924e-06,
+      "loss": 0.3312,
+      "step": 22808
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240120257929722e-06,
+      "loss": 0.313,
+      "step": 22809
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.239413417270041e-06,
+      "loss": 0.3242,
+      "step": 22810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.238706674083763e-06,
+      "loss": 0.3015,
+      "step": 22811
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2380000283797665e-06,
+      "loss": 0.3074,
+      "step": 22812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2372934801669246e-06,
+      "loss": 0.3084,
+      "step": 22813
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.236587029454107e-06,
+      "loss": 0.3305,
+      "step": 22814
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.235880676250195e-06,
+      "loss": 0.308,
+      "step": 22815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2351744205640534e-06,
+      "loss": 0.3271,
+      "step": 22816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2344682624045567e-06,
+      "loss": 0.3372,
+      "step": 22817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.233762201780576e-06,
+      "loss": 0.3165,
+      "step": 22818
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2330562387009745e-06,
+      "loss": 0.3383,
+      "step": 22819
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.232350373174621e-06,
+      "loss": 0.3223,
+      "step": 22820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2316446052103846e-06,
+      "loss": 0.2879,
+      "step": 22821
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230938934817125e-06,
+      "loss": 0.3044,
+      "step": 22822
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23023336200371e-06,
+      "loss": 0.2744,
+      "step": 22823
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.229527886778996e-06,
+      "loss": 0.3021,
+      "step": 22824
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22882250915185e-06,
+      "loss": 0.3319,
+      "step": 22825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.228117229131125e-06,
+      "loss": 0.3437,
+      "step": 22826
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.227412046725683e-06,
+      "loss": 0.3273,
+      "step": 22827
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226706961944385e-06,
+      "loss": 0.2949,
+      "step": 22828
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2260019747960794e-06,
+      "loss": 0.3517,
+      "step": 22829
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.225297085289624e-06,
+      "loss": 0.2981,
+      "step": 22830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224592293433876e-06,
+      "loss": 0.3449,
+      "step": 22831
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2238875992376817e-06,
+      "loss": 0.3038,
+      "step": 22832
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2231830027098967e-06,
+      "loss": 0.2773,
+      "step": 22833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2224785038593687e-06,
+      "loss": 0.3233,
+      "step": 22834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2217741026949447e-06,
+      "loss": 0.2906,
+      "step": 22835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2210697992254726e-06,
+      "loss": 0.3517,
+      "step": 22836
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2203655934598025e-06,
+      "loss": 0.3107,
+      "step": 22837
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2196614854067732e-06,
+      "loss": 0.315,
+      "step": 22838
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.218957475075232e-06,
+      "loss": 0.3013,
+      "step": 22839
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2182535624740233e-06,
+      "loss": 0.326,
+      "step": 22840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2175497476119822e-06,
+      "loss": 0.303,
+      "step": 22841
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.216846030497952e-06,
+      "loss": 0.2987,
+      "step": 22842
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.216142411140776e-06,
+      "loss": 0.3583,
+      "step": 22843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2154388895492864e-06,
+      "loss": 0.3152,
+      "step": 22844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2147354657323163e-06,
+      "loss": 0.3077,
+      "step": 22845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2140321396987087e-06,
+      "loss": 0.3435,
+      "step": 22846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.213328911457291e-06,
+      "loss": 0.3277,
+      "step": 22847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2126257810168983e-06,
+      "loss": 0.2807,
+      "step": 22848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211922748386364e-06,
+      "loss": 0.3212,
+      "step": 22849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2112198135745135e-06,
+      "loss": 0.3209,
+      "step": 22850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.210516976590179e-06,
+      "loss": 0.3076,
+      "step": 22851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.209814237442189e-06,
+      "loss": 0.3136,
+      "step": 22852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2091115961393673e-06,
+      "loss": 0.3573,
+      "step": 22853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.208409052690542e-06,
+      "loss": 0.3519,
+      "step": 22854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207706607104536e-06,
+      "loss": 0.2903,
+      "step": 22855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2070042593901685e-06,
+      "loss": 0.3178,
+      "step": 22856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.206302009556264e-06,
+      "loss": 0.3106,
+      "step": 22857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2055998576116454e-06,
+      "loss": 0.3313,
+      "step": 22858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.204897803565126e-06,
+      "loss": 0.3163,
+      "step": 22859
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2041958474255276e-06,
+      "loss": 0.3161,
+      "step": 22860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.203493989201665e-06,
+      "loss": 0.2914,
+      "step": 22861
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2027922289023574e-06,
+      "loss": 0.3207,
+      "step": 22862
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2020905665364145e-06,
+      "loss": 0.2973,
+      "step": 22863
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2013890021126528e-06,
+      "loss": 0.3164,
+      "step": 22864
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2006875356398815e-06,
+      "loss": 0.2926,
+      "step": 22865
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1999861671269097e-06,
+      "loss": 0.3074,
+      "step": 22866
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1992848965825474e-06,
+      "loss": 0.2975,
+      "step": 22867
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1985837240156073e-06,
+      "loss": 0.3243,
+      "step": 22868
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1978826494348894e-06,
+      "loss": 0.3058,
+      "step": 22869
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1971816728492034e-06,
+      "loss": 0.2781,
+      "step": 22870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1964807942673537e-06,
+      "loss": 0.3194,
+      "step": 22871
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.195780013698141e-06,
+      "loss": 0.3517,
+      "step": 22872
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.195079331150367e-06,
+      "loss": 0.3047,
+      "step": 22873
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.194378746632838e-06,
+      "loss": 0.2822,
+      "step": 22874
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1936782601543482e-06,
+      "loss": 0.3195,
+      "step": 22875
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.192977871723694e-06,
+      "loss": 0.2902,
+      "step": 22876
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1922775813496777e-06,
+      "loss": 0.318,
+      "step": 22877
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1915773890410895e-06,
+      "loss": 0.301,
+      "step": 22878
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190877294806727e-06,
+      "loss": 0.325,
+      "step": 22879
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190177298655386e-06,
+      "loss": 0.304,
+      "step": 22880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.189477400595852e-06,
+      "loss": 0.3408,
+      "step": 22881
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.18877760063692e-06,
+      "loss": 0.3235,
+      "step": 22882
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1880778987873806e-06,
+      "loss": 0.3155,
+      "step": 22883
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1873782950560187e-06,
+      "loss": 0.3312,
+      "step": 22884
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.186678789451625e-06,
+      "loss": 0.3253,
+      "step": 22885
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.18597938198298e-06,
+      "loss": 0.2967,
+      "step": 22886
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1852800726588754e-06,
+      "loss": 0.3393,
+      "step": 22887
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1845808614880872e-06,
+      "loss": 0.3356,
+      "step": 22888
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1838817484794052e-06,
+      "loss": 0.3239,
+      "step": 22889
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.183182733641602e-06,
+      "loss": 0.3272,
+      "step": 22890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1824838169834616e-06,
+      "loss": 0.3189,
+      "step": 22891
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181784998513766e-06,
+      "loss": 0.3148,
+      "step": 22892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1810862782412855e-06,
+      "loss": 0.3227,
+      "step": 22893
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1803876561747984e-06,
+      "loss": 0.3342,
+      "step": 22894
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179689132323084e-06,
+      "loss": 0.3231,
+      "step": 22895
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.178990706694909e-06,
+      "loss": 0.2875,
+      "step": 22896
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1782923792990517e-06,
+      "loss": 0.3191,
+      "step": 22897
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1775941501442775e-06,
+      "loss": 0.2982,
+      "step": 22898
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176896019239361e-06,
+      "loss": 0.321,
+      "step": 22899
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176197986593066e-06,
+      "loss": 0.3174,
+      "step": 22900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1755000522141633e-06,
+      "loss": 0.29,
+      "step": 22901
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1748022161114203e-06,
+      "loss": 0.3202,
+      "step": 22902
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1741044782935973e-06,
+      "loss": 0.3196,
+      "step": 22903
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1734068387694594e-06,
+      "loss": 0.3148,
+      "step": 22904
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.172709297547774e-06,
+      "loss": 0.3143,
+      "step": 22905
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1720118546372946e-06,
+      "loss": 0.331,
+      "step": 22906
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.171314510046788e-06,
+      "loss": 0.3156,
+      "step": 22907
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.17061726378501e-06,
+      "loss": 0.2958,
+      "step": 22908
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.169920115860714e-06,
+      "loss": 0.3058,
+      "step": 22909
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1692230662826595e-06,
+      "loss": 0.3255,
+      "step": 22910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.168526115059606e-06,
+      "loss": 0.3083,
+      "step": 22911
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1678292622002994e-06,
+      "loss": 0.298,
+      "step": 22912
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1671325077134963e-06,
+      "loss": 0.2993,
+      "step": 22913
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1664358516079497e-06,
+      "loss": 0.3072,
+      "step": 22914
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1657392938924047e-06,
+      "loss": 0.2973,
+      "step": 22915
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1650428345756137e-06,
+      "loss": 0.2942,
+      "step": 22916
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1643464736663266e-06,
+      "loss": 0.3005,
+      "step": 22917
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.163650211173286e-06,
+      "loss": 0.3123,
+      "step": 22918
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1629540471052347e-06,
+      "loss": 0.3085,
+      "step": 22919
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1622579814709223e-06,
+      "loss": 0.3409,
+      "step": 22920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.161562014279086e-06,
+      "loss": 0.3304,
+      "step": 22921
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1608661455384705e-06,
+      "loss": 0.2999,
+      "step": 22922
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1601703752578172e-06,
+      "loss": 0.3061,
+      "step": 22923
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.159474703445861e-06,
+      "loss": 0.2826,
+      "step": 22924
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1587791301113402e-06,
+      "loss": 0.2899,
+      "step": 22925
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1580836552629968e-06,
+      "loss": 0.3313,
+      "step": 22926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.157388278909559e-06,
+      "loss": 0.328,
+      "step": 22927
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1566930010597653e-06,
+      "loss": 0.3082,
+      "step": 22928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155997821722348e-06,
+      "loss": 0.3132,
+      "step": 22929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1553027409060344e-06,
+      "loss": 0.3227,
+      "step": 22930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154607758619557e-06,
+      "loss": 0.303,
+      "step": 22931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1539128748716466e-06,
+      "loss": 0.3153,
+      "step": 22932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153218089671032e-06,
+      "loss": 0.3251,
+      "step": 22933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.152523403026435e-06,
+      "loss": 0.3302,
+      "step": 22934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.151828814946584e-06,
+      "loss": 0.3333,
+      "step": 22935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.151134325440205e-06,
+      "loss": 0.3293,
+      "step": 22936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1504399345160154e-06,
+      "loss": 0.2889,
+      "step": 22937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1497456421827432e-06,
+      "loss": 0.3103,
+      "step": 22938
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1490514484491044e-06,
+      "loss": 0.3128,
+      "step": 22939
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.148357353323818e-06,
+      "loss": 0.3733,
+      "step": 22940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1476633568156015e-06,
+      "loss": 0.2906,
+      "step": 22941
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.146969458933177e-06,
+      "loss": 0.3069,
+      "step": 22942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1462756596852518e-06,
+      "loss": 0.3212,
+      "step": 22943
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.145581959080545e-06,
+      "loss": 0.289,
+      "step": 22944
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.144888357127771e-06,
+      "loss": 0.2959,
+      "step": 22945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1441948538356374e-06,
+      "loss": 0.3003,
+      "step": 22946
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1435014492128547e-06,
+      "loss": 0.3272,
+      "step": 22947
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.142808143268138e-06,
+      "loss": 0.274,
+      "step": 22948
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1421149360101913e-06,
+      "loss": 0.3341,
+      "step": 22949
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.141421827447717e-06,
+      "loss": 0.3114,
+      "step": 22950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.140728817589428e-06,
+      "loss": 0.359,
+      "step": 22951
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1400359064440234e-06,
+      "loss": 0.3107,
+      "step": 22952
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1393430940202067e-06,
+      "loss": 0.3011,
+      "step": 22953
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1386503803266845e-06,
+      "loss": 0.2967,
+      "step": 22954
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137957765372151e-06,
+      "loss": 0.328,
+      "step": 22955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1372652491653077e-06,
+      "loss": 0.3237,
+      "step": 22956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.136572831714856e-06,
+      "loss": 0.326,
+      "step": 22957
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.135880513029488e-06,
+      "loss": 0.303,
+      "step": 22958
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1351882931179037e-06,
+      "loss": 0.3373,
+      "step": 22959
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1344961719887914e-06,
+      "loss": 0.3237,
+      "step": 22960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.133804149650851e-06,
+      "loss": 0.3182,
+      "step": 22961
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.133112226112769e-06,
+      "loss": 0.3064,
+      "step": 22962
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1324204013832395e-06,
+      "loss": 0.285,
+      "step": 22963
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131728675470949e-06,
+      "loss": 0.332,
+      "step": 22964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1310370483845853e-06,
+      "loss": 0.3215,
+      "step": 22965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1303455201328416e-06,
+      "loss": 0.3084,
+      "step": 22966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1296540907243947e-06,
+      "loss": 0.2883,
+      "step": 22967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1289627601679343e-06,
+      "loss": 0.3563,
+      "step": 22968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1282715284721443e-06,
+      "loss": 0.2958,
+      "step": 22969
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1275803956457054e-06,
+      "loss": 0.3225,
+      "step": 22970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1268893616972942e-06,
+      "loss": 0.3347,
+      "step": 22971
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.126198426635594e-06,
+      "loss": 0.3016,
+      "step": 22972
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1255075904692855e-06,
+      "loss": 0.3046,
+      "step": 22973
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1248168532070403e-06,
+      "loss": 0.2955,
+      "step": 22974
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1241262148575346e-06,
+      "loss": 0.2654,
+      "step": 22975
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1234356754294493e-06,
+      "loss": 0.3001,
+      "step": 22976
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1227452349314504e-06,
+      "loss": 0.3122,
+      "step": 22977
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.122054893372212e-06,
+      "loss": 0.3322,
+      "step": 22978
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.121364650760408e-06,
+      "loss": 0.3114,
+      "step": 22979
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1206745071047054e-06,
+      "loss": 0.2882,
+      "step": 22980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.119984462413769e-06,
+      "loss": 0.3261,
+      "step": 22981
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.119294516696272e-06,
+      "loss": 0.309,
+      "step": 22982
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1186046699608744e-06,
+      "loss": 0.2969,
+      "step": 22983
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1179149222162443e-06,
+      "loss": 0.3306,
+      "step": 22984
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1172252734710464e-06,
+      "loss": 0.3158,
+      "step": 22985
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116535723733938e-06,
+      "loss": 0.3165,
+      "step": 22986
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.115846273013582e-06,
+      "loss": 0.3544,
+      "step": 22987
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.115156921318642e-06,
+      "loss": 0.2914,
+      "step": 22988
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1144676686577692e-06,
+      "loss": 0.3125,
+      "step": 22989
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1137785150396273e-06,
+      "loss": 0.2858,
+      "step": 22990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.113089460472866e-06,
+      "loss": 0.3472,
+      "step": 22991
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1124005049661457e-06,
+      "loss": 0.306,
+      "step": 22992
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.111711648528113e-06,
+      "loss": 0.2951,
+      "step": 22993
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.111022891167428e-06,
+      "loss": 0.2996,
+      "step": 22994
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1103342328927334e-06,
+      "loss": 0.3276,
+      "step": 22995
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1096456737126834e-06,
+      "loss": 0.3124,
+      "step": 22996
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1089572136359284e-06,
+      "loss": 0.3278,
+      "step": 22997
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1082688526711104e-06,
+      "loss": 0.307,
+      "step": 22998
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1075805908268766e-06,
+      "loss": 0.2935,
+      "step": 22999
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1068924281118763e-06,
+      "loss": 0.3226,
+      "step": 23000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.106204364534745e-06,
+      "loss": 0.2981,
+      "step": 23001
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1055164001041327e-06,
+      "loss": 0.3378,
+      "step": 23002
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1048285348286766e-06,
+      "loss": 0.3049,
+      "step": 23003
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1041407687170134e-06,
+      "loss": 0.2959,
+      "step": 23004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.103453101777785e-06,
+      "loss": 0.3082,
+      "step": 23005
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.102765534019627e-06,
+      "loss": 0.2981,
+      "step": 23006
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10207806545118e-06,
+      "loss": 0.3369,
+      "step": 23007
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1013906960810717e-06,
+      "loss": 0.3201,
+      "step": 23008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100703425917937e-06,
+      "loss": 0.3109,
+      "step": 23009
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100016254970414e-06,
+      "loss": 0.3094,
+      "step": 23010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.099329183247126e-06,
+      "loss": 0.3281,
+      "step": 23011
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0986422107567093e-06,
+      "loss": 0.3084,
+      "step": 23012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0979553375077888e-06,
+      "loss": 0.2816,
+      "step": 23013
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0972685635089883e-06,
+      "loss": 0.3483,
+      "step": 23014
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.096581888768938e-06,
+      "loss": 0.2879,
+      "step": 23015
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0958953132962646e-06,
+      "loss": 0.3168,
+      "step": 23016
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.095208837099586e-06,
+      "loss": 0.3176,
+      "step": 23017
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.094522460187527e-06,
+      "loss": 0.2953,
+      "step": 23018
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0938361825687114e-06,
+      "loss": 0.3361,
+      "step": 23019
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0931500042517526e-06,
+      "loss": 0.3233,
+      "step": 23020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092463925245274e-06,
+      "loss": 0.2722,
+      "step": 23021
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0917779455578923e-06,
+      "loss": 0.2913,
+      "step": 23022
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.091092065198225e-06,
+      "loss": 0.3141,
+      "step": 23023
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0904062841748795e-06,
+      "loss": 0.3298,
+      "step": 23024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.089720602496478e-06,
+      "loss": 0.3251,
+      "step": 23025
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0890350201716248e-06,
+      "loss": 0.2944,
+      "step": 23026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.088349537208936e-06,
+      "loss": 0.3272,
+      "step": 23027
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.087664153617022e-06,
+      "loss": 0.3243,
+      "step": 23028
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0869788694044866e-06,
+      "loss": 0.3127,
+      "step": 23029
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.086293684579941e-06,
+      "loss": 0.337,
+      "step": 23030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0856085991519915e-06,
+      "loss": 0.3028,
+      "step": 23031
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0849236131292383e-06,
+      "loss": 0.3112,
+      "step": 23032
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0842387265202912e-06,
+      "loss": 0.312,
+      "step": 23033
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.083553939333749e-06,
+      "loss": 0.3179,
+      "step": 23034
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08286925157821e-06,
+      "loss": 0.3102,
+      "step": 23035
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0821846632622767e-06,
+      "loss": 0.3091,
+      "step": 23036
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0815001743945495e-06,
+      "loss": 0.2707,
+      "step": 23037
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.080815784983622e-06,
+      "loss": 0.3165,
+      "step": 23038
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08013149503809e-06,
+      "loss": 0.3233,
+      "step": 23039
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.079447304566554e-06,
+      "loss": 0.3437,
+      "step": 23040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0787632135776013e-06,
+      "loss": 0.3114,
+      "step": 23041
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.078079222079825e-06,
+      "loss": 0.2942,
+      "step": 23042
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0773953300818204e-06,
+      "loss": 0.3328,
+      "step": 23043
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0767115375921753e-06,
+      "loss": 0.2716,
+      "step": 23044
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0760278446194728e-06,
+      "loss": 0.2929,
+      "step": 23045
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0753442511723055e-06,
+      "loss": 0.3107,
+      "step": 23046
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0746607572592603e-06,
+      "loss": 0.2987,
+      "step": 23047
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0739773628889173e-06,
+      "loss": 0.3426,
+      "step": 23048
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.073294068069862e-06,
+      "loss": 0.3263,
+      "step": 23049
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0726108728106807e-06,
+      "loss": 0.3151,
+      "step": 23050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0719277771199474e-06,
+      "loss": 0.3181,
+      "step": 23051
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0712447810062463e-06,
+      "loss": 0.2968,
+      "step": 23052
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.070561884478156e-06,
+      "loss": 0.291,
+      "step": 23053
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069879087544254e-06,
+      "loss": 0.3094,
+      "step": 23054
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0691963902131116e-06,
+      "loss": 0.3168,
+      "step": 23055
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0685137924933084e-06,
+      "loss": 0.3143,
+      "step": 23056
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.067831294393415e-06,
+      "loss": 0.3156,
+      "step": 23057
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0671488959220044e-06,
+      "loss": 0.3134,
+      "step": 23058
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.2704,
+      "step": 23059
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.065784397898917e-06,
+      "loss": 0.303,
+      "step": 23060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0651022983643766e-06,
+      "loss": 0.2719,
+      "step": 23061
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0644202984925986e-06,
+      "loss": 0.3229,
+      "step": 23062
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0637383982921433e-06,
+      "loss": 0.3352,
+      "step": 23063
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.06305659777158e-06,
+      "loss": 0.3106,
+      "step": 23064
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0623748969394687e-06,
+      "loss": 0.3124,
+      "step": 23065
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0616932958043755e-06,
+      "loss": 0.2872,
+      "step": 23066
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0610117943748566e-06,
+      "loss": 0.3335,
+      "step": 23067
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0603303926594763e-06,
+      "loss": 0.2779,
+      "step": 23068
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0596490906667887e-06,
+      "loss": 0.3266,
+      "step": 23069
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0589678884053545e-06,
+      "loss": 0.2981,
+      "step": 23070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0582867858837295e-06,
+      "loss": 0.3138,
+      "step": 23071
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0576057831104644e-06,
+      "loss": 0.3039,
+      "step": 23072
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0569248800941166e-06,
+      "loss": 0.3466,
+      "step": 23073
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.056244076843239e-06,
+      "loss": 0.3346,
+      "step": 23074
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.055563373366378e-06,
+      "loss": 0.3097,
+      "step": 23075
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0548827696720885e-06,
+      "loss": 0.33,
+      "step": 23076
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0542022657689177e-06,
+      "loss": 0.3141,
+      "step": 23077
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0535218616654074e-06,
+      "loss": 0.3102,
+      "step": 23078
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0528415573701076e-06,
+      "loss": 0.2727,
+      "step": 23079
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052161352891564e-06,
+      "loss": 0.2987,
+      "step": 23080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0514812482383207e-06,
+      "loss": 0.3112,
+      "step": 23081
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.050801243418916e-06,
+      "loss": 0.315,
+      "step": 23082
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0501213384418926e-06,
+      "loss": 0.2854,
+      "step": 23083
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0494415333157925e-06,
+      "loss": 0.3055,
+      "step": 23084
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0487618280491506e-06,
+      "loss": 0.2818,
+      "step": 23085
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.048082222650507e-06,
+      "loss": 0.2556,
+      "step": 23086
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047402717128397e-06,
+      "loss": 0.2971,
+      "step": 23087
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.046723311491351e-06,
+      "loss": 0.3129,
+      "step": 23088
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0460440057479058e-06,
+      "loss": 0.2872,
+      "step": 23089
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045364799906596e-06,
+      "loss": 0.3255,
+      "step": 23090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0446856939759473e-06,
+      "loss": 0.2645,
+      "step": 23091
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0440066879644915e-06,
+      "loss": 0.3267,
+      "step": 23092
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0433277818807595e-06,
+      "loss": 0.2978,
+      "step": 23093
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0426489757332748e-06,
+      "loss": 0.3238,
+      "step": 23094
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0419702695305642e-06,
+      "loss": 0.3522,
+      "step": 23095
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0412916632811554e-06,
+      "loss": 0.302,
+      "step": 23096
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0406131569935682e-06,
+      "loss": 0.3256,
+      "step": 23097
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0399347506763234e-06,
+      "loss": 0.2993,
+      "step": 23098
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039256444337947e-06,
+      "loss": 0.3172,
+      "step": 23099
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038578237986952e-06,
+      "loss": 0.2899,
+      "step": 23100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0379001316318613e-06,
+      "loss": 0.3118,
+      "step": 23101
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037222125281193e-06,
+      "loss": 0.3135,
+      "step": 23102
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.036544218943458e-06,
+      "loss": 0.3176,
+      "step": 23103
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0358664126271733e-06,
+      "loss": 0.332,
+      "step": 23104
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035188706340856e-06,
+      "loss": 0.309,
+      "step": 23105
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0345111000930108e-06,
+      "loss": 0.2882,
+      "step": 23106
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.033833593892156e-06,
+      "loss": 0.3056,
+      "step": 23107
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0331561877467966e-06,
+      "loss": 0.2961,
+      "step": 23108
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0324788816654396e-06,
+      "loss": 0.3015,
+      "step": 23109
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.031801675656594e-06,
+      "loss": 0.3098,
+      "step": 23110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0311245697287685e-06,
+      "loss": 0.2904,
+      "step": 23111
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.030447563890462e-06,
+      "loss": 0.2868,
+      "step": 23112
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.02977065815018e-06,
+      "loss": 0.2934,
+      "step": 23113
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.029093852516425e-06,
+      "loss": 0.3118,
+      "step": 23114
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.028417146997701e-06,
+      "loss": 0.3115,
+      "step": 23115
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027740541602501e-06,
+      "loss": 0.2953,
+      "step": 23116
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027064036339329e-06,
+      "loss": 0.3205,
+      "step": 23117
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0263876312166786e-06,
+      "loss": 0.3546,
+      "step": 23118
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025711326243044e-06,
+      "loss": 0.2852,
+      "step": 23119
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0250351214269215e-06,
+      "loss": 0.3067,
+      "step": 23120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.024359016776808e-06,
+      "loss": 0.3181,
+      "step": 23121
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0236830123011884e-06,
+      "loss": 0.2788,
+      "step": 23122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0230071080085566e-06,
+      "loss": 0.3049,
+      "step": 23123
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0223313039074055e-06,
+      "loss": 0.3145,
+      "step": 23124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.021655600006218e-06,
+      "loss": 0.3224,
+      "step": 23125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0209799963134823e-06,
+      "loss": 0.3509,
+      "step": 23126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020304492837687e-06,
+      "loss": 0.2934,
+      "step": 23127
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.019629089587315e-06,
+      "loss": 0.3039,
+      "step": 23128
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.018953786570845e-06,
+      "loss": 0.2835,
+      "step": 23129
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0182785837967655e-06,
+      "loss": 0.3411,
+      "step": 23130
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0176034812735525e-06,
+      "loss": 0.3287,
+      "step": 23131
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0169284790096856e-06,
+      "loss": 0.2687,
+      "step": 23132
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.016253577013647e-06,
+      "loss": 0.3238,
+      "step": 23133
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0155787752939073e-06,
+      "loss": 0.3066,
+      "step": 23134
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0149040738589463e-06,
+      "loss": 0.2986,
+      "step": 23135
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0142294727172394e-06,
+      "loss": 0.3097,
+      "step": 23136
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.013554971877255e-06,
+      "loss": 0.2839,
+      "step": 23137
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012880571347471e-06,
+      "loss": 0.3025,
+      "step": 23138
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012206271136353e-06,
+      "loss": 0.3211,
+      "step": 23139
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.01153207125237e-06,
+      "loss": 0.2877,
+      "step": 23140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0108579717039913e-06,
+      "loss": 0.3269,
+      "step": 23141
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0101839724996865e-06,
+      "loss": 0.3376,
+      "step": 23142
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0095100736479154e-06,
+      "loss": 0.3238,
+      "step": 23143
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0088362751571468e-06,
+      "loss": 0.2683,
+      "step": 23144
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0081625770358437e-06,
+      "loss": 0.3071,
+      "step": 23145
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0074889792924633e-06,
+      "loss": 0.3222,
+      "step": 23146
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.006815481935469e-06,
+      "loss": 0.3329,
+      "step": 23147
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0061420849733225e-06,
+      "loss": 0.3139,
+      "step": 23148
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.00546878841448e-06,
+      "loss": 0.3213,
+      "step": 23149
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0047955922673934e-06,
+      "loss": 0.3274,
+      "step": 23150
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0041224965405225e-06,
+      "loss": 0.3094,
+      "step": 23151
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003449501242324e-06,
+      "loss": 0.3194,
+      "step": 23152
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0027766063812446e-06,
+      "loss": 0.312,
+      "step": 23153
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.002103811965739e-06,
+      "loss": 0.2826,
+      "step": 23154
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.00143111800426e-06,
+      "loss": 0.3174,
+      "step": 23155
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0007585245052508e-06,
+      "loss": 0.2955,
+      "step": 23156
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.000086031477164e-06,
+      "loss": 0.2998,
+      "step": 23157
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.999413638928446e-06,
+      "loss": 0.2791,
+      "step": 23158
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9987413468675408e-06,
+      "loss": 0.3125,
+      "step": 23159
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.998069155302891e-06,
+      "loss": 0.3339,
+      "step": 23160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997397064242943e-06,
+      "loss": 0.2862,
+      "step": 23161
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.996725073696133e-06,
+      "loss": 0.3237,
+      "step": 23162
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9960531836709042e-06,
+      "loss": 0.3059,
+      "step": 23163
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9953813941756984e-06,
+      "loss": 0.292,
+      "step": 23164
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994709705218948e-06,
+      "loss": 0.2704,
+      "step": 23165
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9940381168090927e-06,
+      "loss": 0.3064,
+      "step": 23166
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9933666289545683e-06,
+      "loss": 0.3417,
+      "step": 23167
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9926952416638057e-06,
+      "loss": 0.3079,
+      "step": 23168
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9920239549452426e-06,
+      "loss": 0.3178,
+      "step": 23169
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.991352768807303e-06,
+      "loss": 0.3448,
+      "step": 23170
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.990681683258425e-06,
+      "loss": 0.285,
+      "step": 23171
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9900106983070313e-06,
+      "loss": 0.2948,
+      "step": 23172
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9893398139615527e-06,
+      "loss": 0.3029,
+      "step": 23173
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9886690302304135e-06,
+      "loss": 0.2849,
+      "step": 23174
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.98799834712204e-06,
+      "loss": 0.3105,
+      "step": 23175
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9873277646448575e-06,
+      "loss": 0.311,
+      "step": 23176
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986657282807285e-06,
+      "loss": 0.3052,
+      "step": 23177
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9859869016177457e-06,
+      "loss": 0.2845,
+      "step": 23178
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985316621084663e-06,
+      "loss": 0.3121,
+      "step": 23179
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.984646441216449e-06,
+      "loss": 0.2784,
+      "step": 23180
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.983976362021528e-06,
+      "loss": 0.3231,
+      "step": 23181
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.983306383508312e-06,
+      "loss": 0.2621,
+      "step": 23182
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982636505685215e-06,
+      "loss": 0.3235,
+      "step": 23183
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9819667285606526e-06,
+      "loss": 0.3027,
+      "step": 23184
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9812970521430388e-06,
+      "loss": 0.3225,
+      "step": 23185
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.980627476440782e-06,
+      "loss": 0.3322,
+      "step": 23186
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.979958001462291e-06,
+      "loss": 0.2926,
+      "step": 23187
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9792886272159783e-06,
+      "loss": 0.3177,
+      "step": 23188
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9786193537102517e-06,
+      "loss": 0.3078,
+      "step": 23189
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9779501809535127e-06,
+      "loss": 0.3464,
+      "step": 23190
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9772811089541722e-06,
+      "loss": 0.2714,
+      "step": 23191
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9766121377206295e-06,
+      "loss": 0.3264,
+      "step": 23192
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9759432672612855e-06,
+      "loss": 0.3002,
+      "step": 23193
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9752744975845426e-06,
+      "loss": 0.2967,
+      "step": 23194
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9746058286988044e-06,
+      "loss": 0.3216,
+      "step": 23195
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.973937260612464e-06,
+      "loss": 0.3368,
+      "step": 23196
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.973268793333921e-06,
+      "loss": 0.3154,
+      "step": 23197
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972600426871575e-06,
+      "loss": 0.3358,
+      "step": 23198
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.971932161233814e-06,
+      "loss": 0.2985,
+      "step": 23199
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9712639964290335e-06,
+      "loss": 0.3252,
+      "step": 23200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.970595932465631e-06,
+      "loss": 0.3145,
+      "step": 23201
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9699279693519923e-06,
+      "loss": 0.2996,
+      "step": 23202
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.969260107096507e-06,
+      "loss": 0.3304,
+      "step": 23203
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.968592345707565e-06,
+      "loss": 0.3104,
+      "step": 23204
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967924685193552e-06,
+      "loss": 0.3382,
+      "step": 23205
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967257125562854e-06,
+      "loss": 0.2954,
+      "step": 23206
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9665896668238593e-06,
+      "loss": 0.3002,
+      "step": 23207
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9659223089849457e-06,
+      "loss": 0.3294,
+      "step": 23208
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9652550520544977e-06,
+      "loss": 0.2964,
+      "step": 23209
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9645878960408993e-06,
+      "loss": 0.3276,
+      "step": 23210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9639208409525234e-06,
+      "loss": 0.3323,
+      "step": 23211
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9632538867977557e-06,
+      "loss": 0.2986,
+      "step": 23212
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9625870335849685e-06,
+      "loss": 0.3177,
+      "step": 23213
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9619202813225357e-06,
+      "loss": 0.2736,
+      "step": 23214
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9612536300188345e-06,
+      "loss": 0.3295,
+      "step": 23215
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9605870796822403e-06,
+      "loss": 0.3226,
+      "step": 23216
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9599206303211192e-06,
+      "loss": 0.3513,
+      "step": 23217
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9592542819438464e-06,
+      "loss": 0.3202,
+      "step": 23218
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9585880345587925e-06,
+      "loss": 0.326,
+      "step": 23219
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957921888174321e-06,
+      "loss": 0.307,
+      "step": 23220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572558427988008e-06,
+      "loss": 0.3203,
+      "step": 23221
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9565898984406007e-06,
+      "loss": 0.3137,
+      "step": 23222
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9559240551080817e-06,
+      "loss": 0.3153,
+      "step": 23223
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9552583128096035e-06,
+      "loss": 0.2949,
+      "step": 23224
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9545926715535323e-06,
+      "loss": 0.2974,
+      "step": 23225
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9539271313482312e-06,
+      "loss": 0.3257,
+      "step": 23226
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953261692202052e-06,
+      "loss": 0.2903,
+      "step": 23227
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.952596354123357e-06,
+      "loss": 0.2955,
+      "step": 23228
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.951931117120505e-06,
+      "loss": 0.2875,
+      "step": 23229
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9512659812018463e-06,
+      "loss": 0.2999,
+      "step": 23230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.950600946375738e-06,
+      "loss": 0.3219,
+      "step": 23231
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9499360126505352e-06,
+      "loss": 0.3052,
+      "step": 23232
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9492711800345867e-06,
+      "loss": 0.3092,
+      "step": 23233
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94860644853624e-06,
+      "loss": 0.298,
+      "step": 23234
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9479418181638508e-06,
+      "loss": 0.281,
+      "step": 23235
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9472772889257597e-06,
+      "loss": 0.3095,
+      "step": 23236
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9466128608303182e-06,
+      "loss": 0.3338,
+      "step": 23237
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.945948533885872e-06,
+      "loss": 0.3024,
+      "step": 23238
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94528430810076e-06,
+      "loss": 0.291,
+      "step": 23239
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944620183483329e-06,
+      "loss": 0.3244,
+      "step": 23240
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.943956160041922e-06,
+      "loss": 0.3019,
+      "step": 23241
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.943292237784875e-06,
+      "loss": 0.3361,
+      "step": 23242
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94262841672053e-06,
+      "loss": 0.3099,
+      "step": 23243
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.941964696857225e-06,
+      "loss": 0.344,
+      "step": 23244
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9413010782032903e-06,
+      "loss": 0.2878,
+      "step": 23245
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.940637560767067e-06,
+      "loss": 0.3344,
+      "step": 23246
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.939974144556889e-06,
+      "loss": 0.3401,
+      "step": 23247
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.939310829581085e-06,
+      "loss": 0.34,
+      "step": 23248
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9386476158479884e-06,
+      "loss": 0.3254,
+      "step": 23249
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9379845033659306e-06,
+      "loss": 0.3391,
+      "step": 23250
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.937321492143238e-06,
+      "loss": 0.3118,
+      "step": 23251
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9366585821882377e-06,
+      "loss": 0.2949,
+      "step": 23252
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9359957735092603e-06,
+      "loss": 0.3644,
+      "step": 23253
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9353330661146288e-06,
+      "loss": 0.3339,
+      "step": 23254
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.934670460012662e-06,
+      "loss": 0.3375,
+      "step": 23255
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9340079552116886e-06,
+      "loss": 0.317,
+      "step": 23256
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.933345551720024e-06,
+      "loss": 0.3223,
+      "step": 23257
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9326832495459913e-06,
+      "loss": 0.2834,
+      "step": 23258
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.932021048697912e-06,
+      "loss": 0.3039,
+      "step": 23259
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9313589491840966e-06,
+      "loss": 0.2731,
+      "step": 23260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.930696951012866e-06,
+      "loss": 0.2508,
+      "step": 23261
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9300350541925328e-06,
+      "loss": 0.2983,
+      "step": 23262
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.929373258731414e-06,
+      "loss": 0.3085,
+      "step": 23263
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9287115646378186e-06,
+      "loss": 0.3212,
+      "step": 23264
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.928049971920056e-06,
+      "loss": 0.2957,
+      "step": 23265
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.927388480586441e-06,
+      "loss": 0.3081,
+      "step": 23266
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.926727090645275e-06,
+      "loss": 0.3257,
+      "step": 23267
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9260658021048694e-06,
+      "loss": 0.3017,
+      "step": 23268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9254046149735316e-06,
+      "loss": 0.3163,
+      "step": 23269
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924743529259563e-06,
+      "loss": 0.3483,
+      "step": 23270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9240825449712666e-06,
+      "loss": 0.3158,
+      "step": 23271
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9234216621169487e-06,
+      "loss": 0.3532,
+      "step": 23272
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922760880704904e-06,
+      "loss": 0.3361,
+      "step": 23273
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9221002007434375e-06,
+      "loss": 0.3224,
+      "step": 23274
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921439622240844e-06,
+      "loss": 0.3175,
+      "step": 23275
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9207791452054215e-06,
+      "loss": 0.3171,
+      "step": 23276
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9201187696454648e-06,
+      "loss": 0.3057,
+      "step": 23277
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919458495569271e-06,
+      "loss": 0.2951,
+      "step": 23278
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918798322985127e-06,
+      "loss": 0.2831,
+      "step": 23279
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91813825190133e-06,
+      "loss": 0.3287,
+      "step": 23280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9174782823261717e-06,
+      "loss": 0.313,
+      "step": 23281
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916818414267937e-06,
+      "loss": 0.3289,
+      "step": 23282
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9161586477349146e-06,
+      "loss": 0.3097,
+      "step": 23283
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9154989827353953e-06,
+      "loss": 0.3311,
+      "step": 23284
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.914839419277659e-06,
+      "loss": 0.2807,
+      "step": 23285
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9141799573699936e-06,
+      "loss": 0.3073,
+      "step": 23286
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.913520597020683e-06,
+      "loss": 0.31,
+      "step": 23287
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9128613382380024e-06,
+      "loss": 0.3033,
+      "step": 23288
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9122021810302372e-06,
+      "loss": 0.3233,
+      "step": 23289
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.911543125405668e-06,
+      "loss": 0.3159,
+      "step": 23290
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.910884171372568e-06,
+      "loss": 0.315,
+      "step": 23291
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.910225318939215e-06,
+      "loss": 0.3275,
+      "step": 23292
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9095665681138876e-06,
+      "loss": 0.3253,
+      "step": 23293
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9089079189048554e-06,
+      "loss": 0.289,
+      "step": 23294
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.908249371320393e-06,
+      "loss": 0.2917,
+      "step": 23295
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.907590925368773e-06,
+      "loss": 0.2966,
+      "step": 23296
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.906932581058265e-06,
+      "loss": 0.3054,
+      "step": 23297
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.906274338397135e-06,
+      "loss": 0.3397,
+      "step": 23298
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9056161973936516e-06,
+      "loss": 0.321,
+      "step": 23299
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.904958158056086e-06,
+      "loss": 0.2717,
+      "step": 23300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9043002203926951e-06,
+      "loss": 0.3044,
+      "step": 23301
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9036423844117479e-06,
+      "loss": 0.3165,
+      "step": 23302
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9029846501215078e-06,
+      "loss": 0.3126,
+      "step": 23303
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.902327017530232e-06,
+      "loss": 0.2967,
+      "step": 23304
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9016694866461826e-06,
+      "loss": 0.3342,
+      "step": 23305
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9010120574776191e-06,
+      "loss": 0.2777,
+      "step": 23306
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9003547300327996e-06,
+      "loss": 0.2932,
+      "step": 23307
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.899697504319975e-06,
+      "loss": 0.3078,
+      "step": 23308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8990403803474056e-06,
+      "loss": 0.3224,
+      "step": 23309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8983833581233402e-06,
+      "loss": 0.3549,
+      "step": 23310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8977264376560333e-06,
+      "loss": 0.3442,
+      "step": 23311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8970696189537397e-06,
+      "loss": 0.2796,
+      "step": 23312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8964129020247024e-06,
+      "loss": 0.3538,
+      "step": 23313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8957562868771729e-06,
+      "loss": 0.3214,
+      "step": 23314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.895099773519401e-06,
+      "loss": 0.3055,
+      "step": 23315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.894443361959627e-06,
+      "loss": 0.3458,
+      "step": 23316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.893787052206101e-06,
+      "loss": 0.307,
+      "step": 23317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8931308442670638e-06,
+      "loss": 0.319,
+      "step": 23318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8924747381507546e-06,
+      "loss": 0.3001,
+      "step": 23319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.891818733865417e-06,
+      "loss": 0.3209,
+      "step": 23320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.891162831419292e-06,
+      "loss": 0.3294,
+      "step": 23321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8905070308206142e-06,
+      "loss": 0.3198,
+      "step": 23322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8898513320776225e-06,
+      "loss": 0.3327,
+      "step": 23323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8891957351985536e-06,
+      "loss": 0.2967,
+      "step": 23324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8885402401916385e-06,
+      "loss": 0.3146,
+      "step": 23325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8878848470651135e-06,
+      "loss": 0.3093,
+      "step": 23326
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8872295558272103e-06,
+      "loss": 0.3083,
+      "step": 23327
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.886574366486159e-06,
+      "loss": 0.3096,
+      "step": 23328
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8859192790501846e-06,
+      "loss": 0.2918,
+      "step": 23329
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8852642935275223e-06,
+      "loss": 0.3083,
+      "step": 23330
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8846094099263911e-06,
+      "loss": 0.3201,
+      "step": 23331
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8839546282550213e-06,
+      "loss": 0.2922,
+      "step": 23332
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8832999485216351e-06,
+      "loss": 0.3371,
+      "step": 23333
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8826453707344593e-06,
+      "loss": 0.3684,
+      "step": 23334
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8819908949017097e-06,
+      "loss": 0.3136,
+      "step": 23335
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8813365210316082e-06,
+      "loss": 0.2937,
+      "step": 23336
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8806822491323785e-06,
+      "loss": 0.3095,
+      "step": 23337
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.880028079212234e-06,
+      "loss": 0.3156,
+      "step": 23338
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.879374011279388e-06,
+      "loss": 0.2861,
+      "step": 23339
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.878720045342063e-06,
+      "loss": 0.3007,
+      "step": 23340
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8780661814084667e-06,
+      "loss": 0.2901,
+      "step": 23341
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8774124194868126e-06,
+      "loss": 0.2815,
+      "step": 23342
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8767587595853175e-06,
+      "loss": 0.3291,
+      "step": 23343
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8761052017121839e-06,
+      "loss": 0.3105,
+      "step": 23344
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.875451745875625e-06,
+      "loss": 0.3074,
+      "step": 23345
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8747983920838486e-06,
+      "loss": 0.29,
+      "step": 23346
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8741451403450573e-06,
+      "loss": 0.3105,
+      "step": 23347
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.873491990667461e-06,
+      "loss": 0.3195,
+      "step": 23348
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8728389430592576e-06,
+      "loss": 0.2971,
+      "step": 23349
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8721859975286548e-06,
+      "loss": 0.3156,
+      "step": 23350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8715331540838488e-06,
+      "loss": 0.3391,
+      "step": 23351
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8708804127330437e-06,
+      "loss": 0.3253,
+      "step": 23352
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8702277734844331e-06,
+      "loss": 0.286,
+      "step": 23353
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8695752363462171e-06,
+      "loss": 0.3227,
+      "step": 23354
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8689228013265948e-06,
+      "loss": 0.3016,
+      "step": 23355
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.868270468433755e-06,
+      "loss": 0.3178,
+      "step": 23356
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8676182376758922e-06,
+      "loss": 0.3069,
+      "step": 23357
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8669661090612035e-06,
+      "loss": 0.331,
+      "step": 23358
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8663140825978732e-06,
+      "loss": 0.3017,
+      "step": 23359
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.865662158294096e-06,
+      "loss": 0.2968,
+      "step": 23360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8650103361580584e-06,
+      "loss": 0.3524,
+      "step": 23361
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8643586161979432e-06,
+      "loss": 0.2967,
+      "step": 23362
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8637069984219391e-06,
+      "loss": 0.3064,
+      "step": 23363
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.863055482838234e-06,
+      "loss": 0.2807,
+      "step": 23364
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.862404069455005e-06,
+      "loss": 0.3026,
+      "step": 23365
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8617527582804374e-06,
+      "loss": 0.3031,
+      "step": 23366
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8611015493227124e-06,
+      "loss": 0.2687,
+      "step": 23367
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8604504425900049e-06,
+      "loss": 0.3154,
+      "step": 23368
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.859799438090497e-06,
+      "loss": 0.2989,
+      "step": 23369
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8591485358323658e-06,
+      "loss": 0.3091,
+      "step": 23370
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8584977358237855e-06,
+      "loss": 0.3123,
+      "step": 23371
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8578470380729263e-06,
+      "loss": 0.3084,
+      "step": 23372
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8571964425879652e-06,
+      "loss": 0.3203,
+      "step": 23373
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8565459493770744e-06,
+      "loss": 0.3065,
+      "step": 23374
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8558955584484206e-06,
+      "loss": 0.3363,
+      "step": 23375
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8552452698101741e-06,
+      "loss": 0.3032,
+      "step": 23376
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8545950834705074e-06,
+      "loss": 0.2997,
+      "step": 23377
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.853944999437578e-06,
+      "loss": 0.3022,
+      "step": 23378
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8532950177195563e-06,
+      "loss": 0.3189,
+      "step": 23379
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8526451383246092e-06,
+      "loss": 0.2976,
+      "step": 23380
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8519953612608942e-06,
+      "loss": 0.3259,
+      "step": 23381
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.851345686536572e-06,
+      "loss": 0.3144,
+      "step": 23382
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8506961141598057e-06,
+      "loss": 0.3234,
+      "step": 23383
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.850046644138751e-06,
+      "loss": 0.3408,
+      "step": 23384
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8493972764815671e-06,
+      "loss": 0.3088,
+      "step": 23385
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8487480111964119e-06,
+      "loss": 0.2879,
+      "step": 23386
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8480988482914352e-06,
+      "loss": 0.3647,
+      "step": 23387
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8474497877747943e-06,
+      "loss": 0.2885,
+      "step": 23388
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8468008296546436e-06,
+      "loss": 0.3289,
+      "step": 23389
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8461519739391276e-06,
+      "loss": 0.3085,
+      "step": 23390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8455032206364032e-06,
+      "loss": 0.2876,
+      "step": 23391
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8448545697546138e-06,
+      "loss": 0.3206,
+      "step": 23392
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8442060213019052e-06,
+      "loss": 0.3131,
+      "step": 23393
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8435575752864265e-06,
+      "loss": 0.3055,
+      "step": 23394
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8429092317163244e-06,
+      "loss": 0.3073,
+      "step": 23395
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8422609905997358e-06,
+      "loss": 0.2828,
+      "step": 23396
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8416128519448062e-06,
+      "loss": 0.3057,
+      "step": 23397
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8409648157596782e-06,
+      "loss": 0.278,
+      "step": 23398
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8403168820524875e-06,
+      "loss": 0.3199,
+      "step": 23399
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8396690508313731e-06,
+      "loss": 0.3265,
+      "step": 23400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.839021322104475e-06,
+      "loss": 0.3317,
+      "step": 23401
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.838373695879927e-06,
+      "loss": 0.3376,
+      "step": 23402
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.837726172165859e-06,
+      "loss": 0.3126,
+      "step": 23403
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8370787509704102e-06,
+      "loss": 0.319,
+      "step": 23404
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8364314323017074e-06,
+      "loss": 0.3039,
+      "step": 23405
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.835784216167882e-06,
+      "loss": 0.3188,
+      "step": 23406
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8351371025770648e-06,
+      "loss": 0.3283,
+      "step": 23407
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8344900915373854e-06,
+      "loss": 0.3175,
+      "step": 23408
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8338431830569648e-06,
+      "loss": 0.3084,
+      "step": 23409
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.833196377143932e-06,
+      "loss": 0.3085,
+      "step": 23410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.832549673806412e-06,
+      "loss": 0.3366,
+      "step": 23411
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8319030730525256e-06,
+      "loss": 0.2864,
+      "step": 23412
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.831256574890391e-06,
+      "loss": 0.2917,
+      "step": 23413
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.830610179328135e-06,
+      "loss": 0.3536,
+      "step": 23414
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8299638863738688e-06,
+      "loss": 0.3365,
+      "step": 23415
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8293176960357152e-06,
+      "loss": 0.3293,
+      "step": 23416
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8286716083217904e-06,
+      "loss": 0.31,
+      "step": 23417
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.828025623240206e-06,
+      "loss": 0.2759,
+      "step": 23418
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8273797407990768e-06,
+      "loss": 0.2849,
+      "step": 23419
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8267339610065193e-06,
+      "loss": 0.285,
+      "step": 23420
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8260882838706374e-06,
+      "loss": 0.3188,
+      "step": 23421
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8254427093995475e-06,
+      "loss": 0.3318,
+      "step": 23422
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8247972376013556e-06,
+      "loss": 0.3251,
+      "step": 23423
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8241518684841642e-06,
+      "loss": 0.2854,
+      "step": 23424
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8235066020560844e-06,
+      "loss": 0.3119,
+      "step": 23425
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.822861438325222e-06,
+      "loss": 0.3028,
+      "step": 23426
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8222163772996749e-06,
+      "loss": 0.3148,
+      "step": 23427
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8215714189875489e-06,
+      "loss": 0.3462,
+      "step": 23428
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8209265633969454e-06,
+      "loss": 0.3297,
+      "step": 23429
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.820281810535961e-06,
+      "loss": 0.3305,
+      "step": 23430
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.819637160412694e-06,
+      "loss": 0.324,
+      "step": 23431
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8189926130352453e-06,
+      "loss": 0.3293,
+      "step": 23432
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8183481684117076e-06,
+      "loss": 0.3229,
+      "step": 23433
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8177038265501734e-06,
+      "loss": 0.2874,
+      "step": 23434
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8170595874587393e-06,
+      "loss": 0.3422,
+      "step": 23435
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8164154511454924e-06,
+      "loss": 0.2928,
+      "step": 23436
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.815771417618526e-06,
+      "loss": 0.29,
+      "step": 23437
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8151274868859314e-06,
+      "loss": 0.2962,
+      "step": 23438
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8144836589557913e-06,
+      "loss": 0.3177,
+      "step": 23439
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8138399338361956e-06,
+      "loss": 0.2868,
+      "step": 23440
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8131963115352303e-06,
+      "loss": 0.2915,
+      "step": 23441
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8125527920609754e-06,
+      "loss": 0.305,
+      "step": 23442
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811909375421519e-06,
+      "loss": 0.3122,
+      "step": 23443
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8112660616249366e-06,
+      "loss": 0.3105,
+      "step": 23444
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8106228506793145e-06,
+      "loss": 0.3161,
+      "step": 23445
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8099797425927247e-06,
+      "loss": 0.319,
+      "step": 23446
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8093367373732484e-06,
+      "loss": 0.3251,
+      "step": 23447
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.808693835028963e-06,
+      "loss": 0.3275,
+      "step": 23448
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8080510355679403e-06,
+      "loss": 0.3119,
+      "step": 23449
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8074083389982554e-06,
+      "loss": 0.3304,
+      "step": 23450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.806765745327982e-06,
+      "loss": 0.3242,
+      "step": 23451
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8061232545651885e-06,
+      "loss": 0.2893,
+      "step": 23452
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.805480866717948e-06,
+      "loss": 0.2884,
+      "step": 23453
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8048385817943237e-06,
+      "loss": 0.3044,
+      "step": 23454
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8041963998023882e-06,
+      "loss": 0.3308,
+      "step": 23455
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8035543207502038e-06,
+      "loss": 0.3101,
+      "step": 23456
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8029123446458374e-06,
+      "loss": 0.301,
+      "step": 23457
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8022704714973483e-06,
+      "loss": 0.2949,
+      "step": 23458
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.801628701312802e-06,
+      "loss": 0.3124,
+      "step": 23459
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8009870341002611e-06,
+      "loss": 0.3071,
+      "step": 23460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8003454698677791e-06,
+      "loss": 0.3027,
+      "step": 23461
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7997040086234173e-06,
+      "loss": 0.2749,
+      "step": 23462
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.799062650375235e-06,
+      "loss": 0.3303,
+      "step": 23463
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7984213951312833e-06,
+      "loss": 0.3345,
+      "step": 23464
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7977802428996194e-06,
+      "loss": 0.3214,
+      "step": 23465
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7971391936882965e-06,
+      "loss": 0.36,
+      "step": 23466
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7964982475053616e-06,
+      "loss": 0.2999,
+      "step": 23467
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7958574043588672e-06,
+      "loss": 0.2822,
+      "step": 23468
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7952166642568668e-06,
+      "loss": 0.3286,
+      "step": 23469
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.794576027207402e-06,
+      "loss": 0.3061,
+      "step": 23470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7939354932185216e-06,
+      "loss": 0.2919,
+      "step": 23471
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7932950622982737e-06,
+      "loss": 0.3051,
+      "step": 23472
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7926547344546974e-06,
+      "loss": 0.3005,
+      "step": 23473
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7920145096958363e-06,
+      "loss": 0.3209,
+      "step": 23474
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7913743880297363e-06,
+      "loss": 0.3095,
+      "step": 23475
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.790734369464433e-06,
+      "loss": 0.3122,
+      "step": 23476
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7900944540079634e-06,
+      "loss": 0.349,
+      "step": 23477
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.78945464166837e-06,
+      "loss": 0.2996,
+      "step": 23478
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.788814932453683e-06,
+      "loss": 0.336,
+      "step": 23479
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7881753263719404e-06,
+      "loss": 0.284,
+      "step": 23480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7875358234311756e-06,
+      "loss": 0.3081,
+      "step": 23481
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7868964236394237e-06,
+      "loss": 0.3301,
+      "step": 23482
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7862571270047103e-06,
+      "loss": 0.3092,
+      "step": 23483
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7856179335350676e-06,
+      "loss": 0.3209,
+      "step": 23484
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7849788432385262e-06,
+      "loss": 0.3448,
+      "step": 23485
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7843398561231096e-06,
+      "loss": 0.3323,
+      "step": 23486
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7837009721968434e-06,
+      "loss": 0.3303,
+      "step": 23487
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7830621914677548e-06,
+      "loss": 0.3292,
+      "step": 23488
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7824235139438627e-06,
+      "loss": 0.3091,
+      "step": 23489
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.781784939633192e-06,
+      "loss": 0.3216,
+      "step": 23490
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.781146468543765e-06,
+      "loss": 0.3249,
+      "step": 23491
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7805081006835955e-06,
+      "loss": 0.3509,
+      "step": 23492
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7798698360607057e-06,
+      "loss": 0.3193,
+      "step": 23493
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7792316746831128e-06,
+      "loss": 0.3018,
+      "step": 23494
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.778593616558829e-06,
+      "loss": 0.3384,
+      "step": 23495
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7779556616958716e-06,
+      "loss": 0.2872,
+      "step": 23496
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7773178101022514e-06,
+      "loss": 0.3334,
+      "step": 23497
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.776680061785978e-06,
+      "loss": 0.3224,
+      "step": 23498
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7760424167550627e-06,
+      "loss": 0.2547,
+      "step": 23499
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.775404875017519e-06,
+      "loss": 0.3072,
+      "step": 23500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7747674365813473e-06,
+      "loss": 0.2837,
+      "step": 23501
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7741301014545576e-06,
+      "loss": 0.3052,
+      "step": 23502
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.773492869645157e-06,
+      "loss": 0.2894,
+      "step": 23503
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7728557411611447e-06,
+      "loss": 0.3103,
+      "step": 23504
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7722187160105242e-06,
+      "loss": 0.2932,
+      "step": 23505
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7715817942013003e-06,
+      "loss": 0.3075,
+      "step": 23506
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7709449757414698e-06,
+      "loss": 0.3268,
+      "step": 23507
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7703082606390287e-06,
+      "loss": 0.3242,
+      "step": 23508
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7696716489019794e-06,
+      "loss": 0.3397,
+      "step": 23509
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7690351405383133e-06,
+      "loss": 0.2998,
+      "step": 23510
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7683987355560261e-06,
+      "loss": 0.3119,
+      "step": 23511
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7677624339631138e-06,
+      "loss": 0.2631,
+      "step": 23512
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7671262357675645e-06,
+      "loss": 0.3025,
+      "step": 23513
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7664901409773705e-06,
+      "loss": 0.29,
+      "step": 23514
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7658541496005245e-06,
+      "loss": 0.2899,
+      "step": 23515
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7652182616450087e-06,
+      "loss": 0.2798,
+      "step": 23516
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7645824771188157e-06,
+      "loss": 0.3351,
+      "step": 23517
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.763946796029925e-06,
+      "loss": 0.3227,
+      "step": 23518
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7633112183863266e-06,
+      "loss": 0.335,
+      "step": 23519
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7626757441959975e-06,
+      "loss": 0.3182,
+      "step": 23520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7620403734669234e-06,
+      "loss": 0.3069,
+      "step": 23521
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7614051062070848e-06,
+      "loss": 0.3455,
+      "step": 23522
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7607699424244583e-06,
+      "loss": 0.3241,
+      "step": 23523
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7601348821270215e-06,
+      "loss": 0.3356,
+      "step": 23524
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7594999253227552e-06,
+      "loss": 0.3063,
+      "step": 23525
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7588650720196287e-06,
+      "loss": 0.3227,
+      "step": 23526
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7582303222256215e-06,
+      "loss": 0.2836,
+      "step": 23527
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7575956759487023e-06,
+      "loss": 0.282,
+      "step": 23528
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7569611331968418e-06,
+      "loss": 0.2968,
+      "step": 23529
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7563266939780098e-06,
+      "loss": 0.2821,
+      "step": 23530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7556923583001796e-06,
+      "loss": 0.3501,
+      "step": 23531
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7550581261713119e-06,
+      "loss": 0.3083,
+      "step": 23532
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.754423997599376e-06,
+      "loss": 0.3129,
+      "step": 23533
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7537899725923391e-06,
+      "loss": 0.3096,
+      "step": 23534
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7531560511581592e-06,
+      "loss": 0.3079,
+      "step": 23535
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.752522233304802e-06,
+      "loss": 0.2866,
+      "step": 23536
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.751888519040229e-06,
+      "loss": 0.3132,
+      "step": 23537
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7512549083723995e-06,
+      "loss": 0.2938,
+      "step": 23538
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750621401309267e-06,
+      "loss": 0.3158,
+      "step": 23539
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7499879978587942e-06,
+      "loss": 0.3131,
+      "step": 23540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7493546980289312e-06,
+      "loss": 0.3352,
+      "step": 23541
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.748721501827636e-06,
+      "loss": 0.3149,
+      "step": 23542
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7480884092628625e-06,
+      "loss": 0.2792,
+      "step": 23543
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7474554203425586e-06,
+      "loss": 0.3224,
+      "step": 23544
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.746822535074678e-06,
+      "loss": 0.3266,
+      "step": 23545
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7461897534671701e-06,
+      "loss": 0.2896,
+      "step": 23546
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7455570755279782e-06,
+      "loss": 0.3426,
+      "step": 23547
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.744924501265055e-06,
+      "loss": 0.2914,
+      "step": 23548
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7442920306863399e-06,
+      "loss": 0.2682,
+      "step": 23549
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7436596637997817e-06,
+      "loss": 0.3257,
+      "step": 23550
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7430274006133174e-06,
+      "loss": 0.2939,
+      "step": 23551
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7423952411348921e-06,
+      "loss": 0.2847,
+      "step": 23552
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7417631853724471e-06,
+      "loss": 0.3187,
+      "step": 23553
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.741131233333917e-06,
+      "loss": 0.2914,
+      "step": 23554
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.74049938502724e-06,
+      "loss": 0.3272,
+      "step": 23555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7398676404603575e-06,
+      "loss": 0.3354,
+      "step": 23556
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7392359996411968e-06,
+      "loss": 0.3474,
+      "step": 23557
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7386044625776966e-06,
+      "loss": 0.2968,
+      "step": 23558
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7379730292777853e-06,
+      "loss": 0.2859,
+      "step": 23559
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7373416997493974e-06,
+      "loss": 0.3425,
+      "step": 23560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7367104740004582e-06,
+      "loss": 0.3036,
+      "step": 23561
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7360793520389007e-06,
+      "loss": 0.2804,
+      "step": 23562
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7354483338726468e-06,
+      "loss": 0.3211,
+      "step": 23563
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7348174195096247e-06,
+      "loss": 0.3274,
+      "step": 23564
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.734186608957762e-06,
+      "loss": 0.3196,
+      "step": 23565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7335559022249748e-06,
+      "loss": 0.3398,
+      "step": 23566
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7329252993191893e-06,
+      "loss": 0.3228,
+      "step": 23567
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7322948002483265e-06,
+      "loss": 0.2882,
+      "step": 23568
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7316644050203023e-06,
+      "loss": 0.2994,
+      "step": 23569
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7310341136430386e-06,
+      "loss": 0.3429,
+      "step": 23570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.73040392612445e-06,
+      "loss": 0.3045,
+      "step": 23571
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7297738424724486e-06,
+      "loss": 0.3274,
+      "step": 23572
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.729143862694951e-06,
+      "loss": 0.3231,
+      "step": 23573
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7285139867998712e-06,
+      "loss": 0.2883,
+      "step": 23574
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.727884214795117e-06,
+      "loss": 0.2868,
+      "step": 23575
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7272545466886003e-06,
+      "loss": 0.2959,
+      "step": 23576
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.726624982488233e-06,
+      "loss": 0.2766,
+      "step": 23577
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7259955222019165e-06,
+      "loss": 0.3101,
+      "step": 23578
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7253661658375597e-06,
+      "loss": 0.3021,
+      "step": 23579
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7247369134030688e-06,
+      "loss": 0.3136,
+      "step": 23580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7241077649063464e-06,
+      "loss": 0.2957,
+      "step": 23581
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7234787203552916e-06,
+      "loss": 0.3189,
+      "step": 23582
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7228497797578104e-06,
+      "loss": 0.3323,
+      "step": 23583
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7222209431217962e-06,
+      "loss": 0.2704,
+      "step": 23584
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.721592210455152e-06,
+      "loss": 0.3373,
+      "step": 23585
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7209635817657744e-06,
+      "loss": 0.3071,
+      "step": 23586
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7203350570615563e-06,
+      "loss": 0.3437,
+      "step": 23587
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7197066363503923e-06,
+      "loss": 0.3049,
+      "step": 23588
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7190783196401784e-06,
+      "loss": 0.3093,
+      "step": 23589
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.718450106938806e-06,
+      "loss": 0.3049,
+      "step": 23590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7178219982541644e-06,
+      "loss": 0.3154,
+      "step": 23591
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7171939935941396e-06,
+      "loss": 0.3072,
+      "step": 23592
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.716566092966625e-06,
+      "loss": 0.3328,
+      "step": 23593
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7159382963795012e-06,
+      "loss": 0.296,
+      "step": 23594
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7153106038406564e-06,
+      "loss": 0.3322,
+      "step": 23595
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7146830153579775e-06,
+      "loss": 0.3434,
+      "step": 23596
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7140555309393403e-06,
+      "loss": 0.3339,
+      "step": 23597
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7134281505926309e-06,
+      "loss": 0.2689,
+      "step": 23598
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7128008743257307e-06,
+      "loss": 0.3275,
+      "step": 23599
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7121737021465123e-06,
+      "loss": 0.3249,
+      "step": 23600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7115466340628583e-06,
+      "loss": 0.2742,
+      "step": 23601
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7109196700826448e-06,
+      "loss": 0.3111,
+      "step": 23602
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7102928102137406e-06,
+      "loss": 0.3144,
+      "step": 23603
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.709666054464023e-06,
+      "loss": 0.2805,
+      "step": 23604
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.709039402841367e-06,
+      "loss": 0.3404,
+      "step": 23605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7084128553536373e-06,
+      "loss": 0.2974,
+      "step": 23606
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7077864120087073e-06,
+      "loss": 0.3083,
+      "step": 23607
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7071600728144467e-06,
+      "loss": 0.3236,
+      "step": 23608
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7065338377787167e-06,
+      "loss": 0.3124,
+      "step": 23609
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7059077069093865e-06,
+      "loss": 0.3143,
+      "step": 23610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7052816802143235e-06,
+      "loss": 0.3052,
+      "step": 23611
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7046557577013868e-06,
+      "loss": 0.3191,
+      "step": 23612
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7040299393784355e-06,
+      "loss": 0.3057,
+      "step": 23613
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7034042252533345e-06,
+      "loss": 0.3233,
+      "step": 23614
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7027786153339398e-06,
+      "loss": 0.3218,
+      "step": 23615
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7021531096281097e-06,
+      "loss": 0.3231,
+      "step": 23616
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7015277081437043e-06,
+      "loss": 0.2978,
+      "step": 23617
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.700902410888573e-06,
+      "loss": 0.2953,
+      "step": 23618
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7002772178705717e-06,
+      "loss": 0.3088,
+      "step": 23619
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6996521290975565e-06,
+      "loss": 0.2887,
+      "step": 23620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6990271445773732e-06,
+      "loss": 0.2828,
+      "step": 23621
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6984022643178754e-06,
+      "loss": 0.3008,
+      "step": 23622
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6977774883269083e-06,
+      "loss": 0.3023,
+      "step": 23623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6971528166123241e-06,
+      "loss": 0.2734,
+      "step": 23624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6965282491819612e-06,
+      "loss": 0.2945,
+      "step": 23625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.69590378604367e-06,
+      "loss": 0.3111,
+      "step": 23626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6952794272052942e-06,
+      "loss": 0.286,
+      "step": 23627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6946551726746718e-06,
+      "loss": 0.3275,
+      "step": 23628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6940310224596457e-06,
+      "loss": 0.3003,
+      "step": 23629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6934069765680572e-06,
+      "loss": 0.3315,
+      "step": 23630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.69278303500774e-06,
+      "loss": 0.2843,
+      "step": 23631
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.692159197786536e-06,
+      "loss": 0.2953,
+      "step": 23632
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6915354649122739e-06,
+      "loss": 0.2978,
+      "step": 23633
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6909118363927956e-06,
+      "loss": 0.2982,
+      "step": 23634
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.690288312235927e-06,
+      "loss": 0.3276,
+      "step": 23635
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6896648924495052e-06,
+      "loss": 0.3299,
+      "step": 23636
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6890415770413549e-06,
+      "loss": 0.3401,
+      "step": 23637
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6884183660193076e-06,
+      "loss": 0.3133,
+      "step": 23638
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687795259391195e-06,
+      "loss": 0.312,
+      "step": 23639
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6871722571648364e-06,
+      "loss": 0.3031,
+      "step": 23640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6865493593480586e-06,
+      "loss": 0.3245,
+      "step": 23641
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6859265659486902e-06,
+      "loss": 0.3057,
+      "step": 23642
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.2745,
+      "step": 23643
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6846812924334554e-06,
+      "loss": 0.2782,
+      "step": 23644
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6840588123332325e-06,
+      "loss": 0.3028,
+      "step": 23645
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6834364366816935e-06,
+      "loss": 0.3249,
+      "step": 23646
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6828141654866592e-06,
+      "loss": 0.3351,
+      "step": 23647
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6821919987559476e-06,
+      "loss": 0.2877,
+      "step": 23648
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.681569936497367e-06,
+      "loss": 0.3022,
+      "step": 23649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6809479787187344e-06,
+      "loss": 0.3361,
+      "step": 23650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6803261254278635e-06,
+      "loss": 0.3416,
+      "step": 23651
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6797043766325604e-06,
+      "loss": 0.3059,
+      "step": 23652
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6790827323406379e-06,
+      "loss": 0.2739,
+      "step": 23653
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6784611925599004e-06,
+      "loss": 0.3189,
+      "step": 23654
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6778397572981585e-06,
+      "loss": 0.2823,
+      "step": 23655
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6772184265632142e-06,
+      "loss": 0.336,
+      "step": 23656
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.676597200362874e-06,
+      "loss": 0.3093,
+      "step": 23657
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6759760787049361e-06,
+      "loss": 0.338,
+      "step": 23658
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6753550615972059e-06,
+      "loss": 0.295,
+      "step": 23659
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6747341490474844e-06,
+      "loss": 0.3048,
+      "step": 23660
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6741133410635658e-06,
+      "loss": 0.3287,
+      "step": 23661
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.673492637653249e-06,
+      "loss": 0.2806,
+      "step": 23662
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6728720388243347e-06,
+      "loss": 0.3132,
+      "step": 23663
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.672251544584611e-06,
+      "loss": 0.3142,
+      "step": 23664
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6716311549418773e-06,
+      "loss": 0.2952,
+      "step": 23665
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6710108699039196e-06,
+      "loss": 0.3067,
+      "step": 23666
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6703906894785348e-06,
+      "loss": 0.3211,
+      "step": 23667
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6697706136735059e-06,
+      "loss": 0.2964,
+      "step": 23668
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6691506424966264e-06,
+      "loss": 0.2849,
+      "step": 23669
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6685307759556824e-06,
+      "loss": 0.2766,
+      "step": 23670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6679110140584564e-06,
+      "loss": 0.2922,
+      "step": 23671
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6672913568127358e-06,
+      "loss": 0.3015,
+      "step": 23672
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6666718042263031e-06,
+      "loss": 0.3253,
+      "step": 23673
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6660523563069387e-06,
+      "loss": 0.3054,
+      "step": 23674
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6654330130624242e-06,
+      "loss": 0.2989,
+      "step": 23675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6648137745005388e-06,
+      "loss": 0.3056,
+      "step": 23676
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6641946406290566e-06,
+      "loss": 0.3037,
+      "step": 23677
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6635756114557567e-06,
+      "loss": 0.308,
+      "step": 23678
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.662956686988416e-06,
+      "loss": 0.2939,
+      "step": 23679
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6623378672348034e-06,
+      "loss": 0.3111,
+      "step": 23680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6617191522026953e-06,
+      "loss": 0.2807,
+      "step": 23681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.661100541899864e-06,
+      "loss": 0.3281,
+      "step": 23682
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6604820363340735e-06,
+      "loss": 0.3492,
+      "step": 23683
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.659863635513096e-06,
+      "loss": 0.2857,
+      "step": 23684
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6592453394447016e-06,
+      "loss": 0.3113,
+      "step": 23685
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6586271481366524e-06,
+      "loss": 0.3049,
+      "step": 23686
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6580090615967103e-06,
+      "loss": 0.3048,
+      "step": 23687
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6573910798326454e-06,
+      "loss": 0.3352,
+      "step": 23688
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.656773202852212e-06,
+      "loss": 0.2916,
+      "step": 23689
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6561554306631745e-06,
+      "loss": 0.3081,
+      "step": 23690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.655537763273295e-06,
+      "loss": 0.328,
+      "step": 23691
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6549202006903264e-06,
+      "loss": 0.2667,
+      "step": 23692
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.654302742922027e-06,
+      "loss": 0.3326,
+      "step": 23693
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.653685389976155e-06,
+      "loss": 0.3201,
+      "step": 23694
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.65306814186046e-06,
+      "loss": 0.286,
+      "step": 23695
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6524509985826987e-06,
+      "loss": 0.2846,
+      "step": 23696
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6518339601506206e-06,
+      "loss": 0.3175,
+      "step": 23697
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6512170265719729e-06,
+      "loss": 0.3009,
+      "step": 23698
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.650600197854506e-06,
+      "loss": 0.3017,
+      "step": 23699
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6499834740059695e-06,
+      "loss": 0.3327,
+      "step": 23700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6493668550341102e-06,
+      "loss": 0.3177,
+      "step": 23701
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6487503409466686e-06,
+      "loss": 0.325,
+      "step": 23702
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6481339317513912e-06,
+      "loss": 0.3133,
+      "step": 23703
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.64751762745602e-06,
+      "loss": 0.292,
+      "step": 23704
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6469014280682939e-06,
+      "loss": 0.3321,
+      "step": 23705
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6462853335959561e-06,
+      "loss": 0.2708,
+      "step": 23706
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.645669344046742e-06,
+      "loss": 0.3133,
+      "step": 23707
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6450534594283873e-06,
+      "loss": 0.2913,
+      "step": 23708
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6444376797486283e-06,
+      "loss": 0.3277,
+      "step": 23709
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.643822005015202e-06,
+      "loss": 0.3261,
+      "step": 23710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6432064352358378e-06,
+      "loss": 0.3273,
+      "step": 23711
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6425909704182685e-06,
+      "loss": 0.2962,
+      "step": 23712
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6419756105702268e-06,
+      "loss": 0.3104,
+      "step": 23713
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6413603556994374e-06,
+      "loss": 0.3178,
+      "step": 23714
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6407452058136298e-06,
+      "loss": 0.3535,
+      "step": 23715
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6401301609205333e-06,
+      "loss": 0.2966,
+      "step": 23716
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6395152210278697e-06,
+      "loss": 0.3346,
+      "step": 23717
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6389003861433616e-06,
+      "loss": 0.3224,
+      "step": 23718
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.638285656274735e-06,
+      "loss": 0.3269,
+      "step": 23719
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6376710314297062e-06,
+      "loss": 0.3224,
+      "step": 23720
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6370565116159976e-06,
+      "loss": 0.3199,
+      "step": 23721
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6364420968413309e-06,
+      "loss": 0.3098,
+      "step": 23722
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6358277871134164e-06,
+      "loss": 0.3141,
+      "step": 23723
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.635213582439974e-06,
+      "loss": 0.3058,
+      "step": 23724
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6345994828287193e-06,
+      "loss": 0.2875,
+      "step": 23725
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6339854882873606e-06,
+      "loss": 0.3375,
+      "step": 23726
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6333715988236165e-06,
+      "loss": 0.3084,
+      "step": 23727
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6327578144451905e-06,
+      "loss": 0.349,
+      "step": 23728
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6321441351597978e-06,
+      "loss": 0.2968,
+      "step": 23729
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6315305609751408e-06,
+      "loss": 0.2688,
+      "step": 23730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6309170918989315e-06,
+      "loss": 0.3291,
+      "step": 23731
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6303037279388688e-06,
+      "loss": 0.2886,
+      "step": 23732
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6296904691026605e-06,
+      "loss": 0.2759,
+      "step": 23733
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.629077315398012e-06,
+      "loss": 0.2932,
+      "step": 23734
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6284642668326179e-06,
+      "loss": 0.3346,
+      "step": 23735
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6278513234141814e-06,
+      "loss": 0.2978,
+      "step": 23736
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6272384851504043e-06,
+      "loss": 0.327,
+      "step": 23737
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.626625752048978e-06,
+      "loss": 0.3195,
+      "step": 23738
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6260131241176037e-06,
+      "loss": 0.3077,
+      "step": 23739
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.62540060136397e-06,
+      "loss": 0.3009,
+      "step": 23740
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.624788183795778e-06,
+      "loss": 0.262,
+      "step": 23741
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6241758714207112e-06,
+      "loss": 0.3216,
+      "step": 23742
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6235636642464657e-06,
+      "loss": 0.3327,
+      "step": 23743
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6229515622807324e-06,
+      "loss": 0.3489,
+      "step": 23744
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6223395655311936e-06,
+      "loss": 0.29,
+      "step": 23745
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.62172767400554e-06,
+      "loss": 0.3388,
+      "step": 23746
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6211158877114574e-06,
+      "loss": 0.2936,
+      "step": 23747
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6205042066566267e-06,
+      "loss": 0.3041,
+      "step": 23748
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6198926308487339e-06,
+      "loss": 0.3255,
+      "step": 23749
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6192811602954605e-06,
+      "loss": 0.3022,
+      "step": 23750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6186697950044817e-06,
+      "loss": 0.3206,
+      "step": 23751
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.61805853498348e-06,
+      "loss": 0.2932,
+      "step": 23752
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6174473802401348e-06,
+      "loss": 0.3294,
+      "step": 23753
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6168363307821178e-06,
+      "loss": 0.3321,
+      "step": 23754
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.616225386617105e-06,
+      "loss": 0.2998,
+      "step": 23755
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.615614547752774e-06,
+      "loss": 0.3053,
+      "step": 23756
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6150038141967917e-06,
+      "loss": 0.2723,
+      "step": 23757
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6143931859568308e-06,
+      "loss": 0.2942,
+      "step": 23758
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.613782663040564e-06,
+      "loss": 0.3262,
+      "step": 23759
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6131722454556564e-06,
+      "loss": 0.3058,
+      "step": 23760
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6125619332097731e-06,
+      "loss": 0.3128,
+      "step": 23761
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6119517263105833e-06,
+      "loss": 0.3024,
+      "step": 23762
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6113416247657476e-06,
+      "loss": 0.3015,
+      "step": 23763
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.610731628582931e-06,
+      "loss": 0.3235,
+      "step": 23764
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6101217377697965e-06,
+      "loss": 0.321,
+      "step": 23765
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.609511952334002e-06,
+      "loss": 0.3163,
+      "step": 23766
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.608902272283206e-06,
+      "loss": 0.2997,
+      "step": 23767
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.60829269762507e-06,
+      "loss": 0.3113,
+      "step": 23768
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6076832283672462e-06,
+      "loss": 0.3259,
+      "step": 23769
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6070738645173922e-06,
+      "loss": 0.3103,
+      "step": 23770
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.606464606083159e-06,
+      "loss": 0.2944,
+      "step": 23771
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6058554530722026e-06,
+      "loss": 0.2991,
+      "step": 23772
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6052464054921691e-06,
+      "loss": 0.3287,
+      "step": 23773
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6046374633507122e-06,
+      "loss": 0.2873,
+      "step": 23774
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6040286266554795e-06,
+      "loss": 0.3297,
+      "step": 23775
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6034198954141168e-06,
+      "loss": 0.3332,
+      "step": 23776
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6028112696342691e-06,
+      "loss": 0.2886,
+      "step": 23777
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6022027493235848e-06,
+      "loss": 0.2976,
+      "step": 23778
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6015943344897022e-06,
+      "loss": 0.3388,
+      "step": 23779
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6009860251402676e-06,
+      "loss": 0.2886,
+      "step": 23780
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.600377821282919e-06,
+      "loss": 0.3203,
+      "step": 23781
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5997697229252928e-06,
+      "loss": 0.321,
+      "step": 23782
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5991617300750295e-06,
+      "loss": 0.3043,
+      "step": 23783
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5985538427397684e-06,
+      "loss": 0.3239,
+      "step": 23784
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.597946060927138e-06,
+      "loss": 0.2939,
+      "step": 23785
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5973383846447766e-06,
+      "loss": 0.3116,
+      "step": 23786
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5967308139003168e-06,
+      "loss": 0.2878,
+      "step": 23787
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5961233487013872e-06,
+      "loss": 0.2778,
+      "step": 23788
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5955159890556182e-06,
+      "loss": 0.3207,
+      "step": 23789
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5949087349706427e-06,
+      "loss": 0.3504,
+      "step": 23790
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5943015864540833e-06,
+      "loss": 0.3151,
+      "step": 23791
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5936945435135653e-06,
+      "loss": 0.316,
+      "step": 23792
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5930876061567157e-06,
+      "loss": 0.2863,
+      "step": 23793
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5924807743911553e-06,
+      "loss": 0.3179,
+      "step": 23794
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5918740482245066e-06,
+      "loss": 0.2756,
+      "step": 23795
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5912674276643936e-06,
+      "loss": 0.3274,
+      "step": 23796
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5906609127184302e-06,
+      "loss": 0.3228,
+      "step": 23797
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590054503394236e-06,
+      "loss": 0.3125,
+      "step": 23798
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5894481996994315e-06,
+      "loss": 0.3134,
+      "step": 23799
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5888420016416261e-06,
+      "loss": 0.273,
+      "step": 23800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5882359092284382e-06,
+      "loss": 0.2822,
+      "step": 23801
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5876299224674785e-06,
+      "loss": 0.3065,
+      "step": 23802
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5870240413663562e-06,
+      "loss": 0.2979,
+      "step": 23803
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5864182659326833e-06,
+      "loss": 0.3309,
+      "step": 23804
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.58581259617407e-06,
+      "loss": 0.2941,
+      "step": 23805
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5852070320981194e-06,
+      "loss": 0.2902,
+      "step": 23806
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5846015737124398e-06,
+      "loss": 0.3317,
+      "step": 23807
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.583996221024635e-06,
+      "loss": 0.3261,
+      "step": 23808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5833909740423125e-06,
+      "loss": 0.3002,
+      "step": 23809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5827858327730683e-06,
+      "loss": 0.3165,
+      "step": 23810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5821807972245073e-06,
+      "loss": 0.3167,
+      "step": 23811
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.581575867404227e-06,
+      "loss": 0.2684,
+      "step": 23812
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5809710433198234e-06,
+      "loss": 0.2937,
+      "step": 23813
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5803663249788948e-06,
+      "loss": 0.3118,
+      "step": 23814
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5797617123890396e-06,
+      "loss": 0.3333,
+      "step": 23815
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5791572055578452e-06,
+      "loss": 0.3253,
+      "step": 23816
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.578552804492909e-06,
+      "loss": 0.3233,
+      "step": 23817
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5779485092018232e-06,
+      "loss": 0.2815,
+      "step": 23818
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5773443196921734e-06,
+      "loss": 0.2737,
+      "step": 23819
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.576740235971551e-06,
+      "loss": 0.2715,
+      "step": 23820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5761362580475449e-06,
+      "loss": 0.3212,
+      "step": 23821
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5755323859277394e-06,
+      "loss": 0.2899,
+      "step": 23822
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5749286196197167e-06,
+      "loss": 0.2936,
+      "step": 23823
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.574324959131065e-06,
+      "loss": 0.2963,
+      "step": 23824
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5737214044693604e-06,
+      "loss": 0.2884,
+      "step": 23825
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5731179556421872e-06,
+      "loss": 0.305,
+      "step": 23826
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5725146126571278e-06,
+      "loss": 0.3397,
+      "step": 23827
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.571911375521753e-06,
+      "loss": 0.282,
+      "step": 23828
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5713082442436445e-06,
+      "loss": 0.2896,
+      "step": 23829
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5707052188303785e-06,
+      "loss": 0.3303,
+      "step": 23830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5701022992895243e-06,
+      "loss": 0.3187,
+      "step": 23831
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5694994856286606e-06,
+      "loss": 0.2808,
+      "step": 23832
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.568896777855352e-06,
+      "loss": 0.3129,
+      "step": 23833
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5682941759771754e-06,
+      "loss": 0.3175,
+      "step": 23834
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.567691680001694e-06,
+      "loss": 0.3351,
+      "step": 23835
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5670892899364788e-06,
+      "loss": 0.3134,
+      "step": 23836
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5664870057890924e-06,
+      "loss": 0.3076,
+      "step": 23837
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5658848275671013e-06,
+      "loss": 0.3408,
+      "step": 23838
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5652827552780725e-06,
+      "loss": 0.3017,
+      "step": 23839
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5646807889295613e-06,
+      "loss": 0.3266,
+      "step": 23840
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5640789285291324e-06,
+      "loss": 0.3245,
+      "step": 23841
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5634771740843468e-06,
+      "loss": 0.3128,
+      "step": 23842
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5628755256027573e-06,
+      "loss": 0.2942,
+      "step": 23843
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5622739830919277e-06,
+      "loss": 0.2717,
+      "step": 23844
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.561672546559405e-06,
+      "loss": 0.3238,
+      "step": 23845
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5610712160127516e-06,
+      "loss": 0.3184,
+      "step": 23846
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5604699914595135e-06,
+      "loss": 0.3257,
+      "step": 23847
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5598688729072454e-06,
+      "loss": 0.3043,
+      "step": 23848
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5592678603634992e-06,
+      "loss": 0.32,
+      "step": 23849
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.558666953835819e-06,
+      "loss": 0.3352,
+      "step": 23850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5580661533317543e-06,
+      "loss": 0.2765,
+      "step": 23851
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5574654588588533e-06,
+      "loss": 0.3334,
+      "step": 23852
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.556864870424658e-06,
+      "loss": 0.329,
+      "step": 23853
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5562643880367145e-06,
+      "loss": 0.3079,
+      "step": 23854
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5556640117025634e-06,
+      "loss": 0.2948,
+      "step": 23855
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.555063741429742e-06,
+      "loss": 0.2876,
+      "step": 23856
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5544635772257942e-06,
+      "loss": 0.3255,
+      "step": 23857
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5538635190982588e-06,
+      "loss": 0.326,
+      "step": 23858
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.553263567054668e-06,
+      "loss": 0.3153,
+      "step": 23859
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5526637211025598e-06,
+      "loss": 0.2731,
+      "step": 23860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.552063981249471e-06,
+      "loss": 0.2822,
+      "step": 23861
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.551464347502929e-06,
+      "loss": 0.3283,
+      "step": 23862
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.550864819870469e-06,
+      "loss": 0.3346,
+      "step": 23863
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5502653983596217e-06,
+      "loss": 0.3094,
+      "step": 23864
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5496660829779143e-06,
+      "loss": 0.3109,
+      "step": 23865
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5490668737328719e-06,
+      "loss": 0.2972,
+      "step": 23866
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5484677706320239e-06,
+      "loss": 0.3215,
+      "step": 23867
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5478687736828934e-06,
+      "loss": 0.3241,
+      "step": 23868
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5472698828930044e-06,
+      "loss": 0.3046,
+      "step": 23869
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5466710982698807e-06,
+      "loss": 0.3172,
+      "step": 23870
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5460724198210397e-06,
+      "loss": 0.34,
+      "step": 23871
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.545473847554002e-06,
+      "loss": 0.2986,
+      "step": 23872
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5448753814762885e-06,
+      "loss": 0.3246,
+      "step": 23873
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5442770215954118e-06,
+      "loss": 0.2999,
+      "step": 23874
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5436787679188913e-06,
+      "loss": 0.3019,
+      "step": 23875
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5430806204542404e-06,
+      "loss": 0.325,
+      "step": 23876
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5424825792089671e-06,
+      "loss": 0.3427,
+      "step": 23877
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5418846441905854e-06,
+      "loss": 0.2886,
+      "step": 23878
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5412868154066096e-06,
+      "loss": 0.2998,
+      "step": 23879
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5406890928645434e-06,
+      "loss": 0.3176,
+      "step": 23880
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5400914765718956e-06,
+      "loss": 0.2867,
+      "step": 23881
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5394939665361718e-06,
+      "loss": 0.287,
+      "step": 23882
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5388965627648811e-06,
+      "loss": 0.3317,
+      "step": 23883
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5382992652655204e-06,
+      "loss": 0.308,
+      "step": 23884
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5377020740455973e-06,
+      "loss": 0.3159,
+      "step": 23885
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.53710498911261e-06,
+      "loss": 0.3077,
+      "step": 23886
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.536508010474056e-06,
+      "loss": 0.3108,
+      "step": 23887
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5359111381374348e-06,
+      "loss": 0.3413,
+      "step": 23888
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.535314372110247e-06,
+      "loss": 0.3159,
+      "step": 23889
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5347177123999823e-06,
+      "loss": 0.3211,
+      "step": 23890
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.534121159014137e-06,
+      "loss": 0.3015,
+      "step": 23891
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5335247119602058e-06,
+      "loss": 0.2775,
+      "step": 23892
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5329283712456767e-06,
+      "loss": 0.3162,
+      "step": 23893
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5323321368780407e-06,
+      "loss": 0.3328,
+      "step": 23894
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5317360088647904e-06,
+      "loss": 0.3051,
+      "step": 23895
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5311399872134103e-06,
+      "loss": 0.3012,
+      "step": 23896
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5305440719313835e-06,
+      "loss": 0.3118,
+      "step": 23897
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5299482630261998e-06,
+      "loss": 0.3017,
+      "step": 23898
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5293525605053384e-06,
+      "loss": 0.3167,
+      "step": 23899
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5287569643762834e-06,
+      "loss": 0.3211,
+      "step": 23900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5281614746465168e-06,
+      "loss": 0.3261,
+      "step": 23901
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5275660913235146e-06,
+      "loss": 0.306,
+      "step": 23902
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5269708144147566e-06,
+      "loss": 0.2699,
+      "step": 23903
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5263756439277221e-06,
+      "loss": 0.3004,
+      "step": 23904
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5257805798698821e-06,
+      "loss": 0.2765,
+      "step": 23905
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5251856222487149e-06,
+      "loss": 0.3088,
+      "step": 23906
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5245907710716912e-06,
+      "loss": 0.3253,
+      "step": 23907
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5239960263462795e-06,
+      "loss": 0.3292,
+      "step": 23908
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5234013880799526e-06,
+      "loss": 0.31,
+      "step": 23909
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5228068562801813e-06,
+      "loss": 0.3102,
+      "step": 23910
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5222124309544272e-06,
+      "loss": 0.3376,
+      "step": 23911
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5216181121101613e-06,
+      "loss": 0.3491,
+      "step": 23912
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5210238997548477e-06,
+      "loss": 0.3089,
+      "step": 23913
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5204297938959479e-06,
+      "loss": 0.2743,
+      "step": 23914
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5198357945409236e-06,
+      "loss": 0.2718,
+      "step": 23915
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5192419016972403e-06,
+      "loss": 0.3136,
+      "step": 23916
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.518648115372352e-06,
+      "loss": 0.3214,
+      "step": 23917
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.518054435573717e-06,
+      "loss": 0.2831,
+      "step": 23918
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.517460862308794e-06,
+      "loss": 0.309,
+      "step": 23919
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.516867395585039e-06,
+      "loss": 0.2659,
+      "step": 23920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5162740354099027e-06,
+      "loss": 0.3122,
+      "step": 23921
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5156807817908393e-06,
+      "loss": 0.3282,
+      "step": 23922
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5150876347353027e-06,
+      "loss": 0.3141,
+      "step": 23923
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5144945942507395e-06,
+      "loss": 0.3155,
+      "step": 23924
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5139016603445977e-06,
+      "loss": 0.3205,
+      "step": 23925
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5133088330243295e-06,
+      "loss": 0.3185,
+      "step": 23926
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5127161122973777e-06,
+      "loss": 0.3113,
+      "step": 23927
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5121234981711851e-06,
+      "loss": 0.275,
+      "step": 23928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5115309906531984e-06,
+      "loss": 0.2831,
+      "step": 23929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5109385897508554e-06,
+      "loss": 0.2808,
+      "step": 23930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5103462954715986e-06,
+      "loss": 0.3191,
+      "step": 23931
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5097541078228718e-06,
+      "loss": 0.2862,
+      "step": 23932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5091620268121055e-06,
+      "loss": 0.2755,
+      "step": 23933
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5085700524467394e-06,
+      "loss": 0.2994,
+      "step": 23934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5079781847342122e-06,
+      "loss": 0.3099,
+      "step": 23935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5073864236819513e-06,
+      "loss": 0.3491,
+      "step": 23936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5067947692973961e-06,
+      "loss": 0.2865,
+      "step": 23937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5062032215879707e-06,
+      "loss": 0.3028,
+      "step": 23938
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5056117805611115e-06,
+      "loss": 0.3131,
+      "step": 23939
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5050204462242423e-06,
+      "loss": 0.3024,
+      "step": 23940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.504429218584793e-06,
+      "loss": 0.3092,
+      "step": 23941
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5038380976501877e-06,
+      "loss": 0.3236,
+      "step": 23942
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5032470834278512e-06,
+      "loss": 0.3463,
+      "step": 23943
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.502656175925209e-06,
+      "loss": 0.3106,
+      "step": 23944
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5020653751496795e-06,
+      "loss": 0.3057,
+      "step": 23945
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5014746811086844e-06,
+      "loss": 0.3139,
+      "step": 23946
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5008840938096448e-06,
+      "loss": 0.2643,
+      "step": 23947
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5002936132599755e-06,
+      "loss": 0.3302,
+      "step": 23948
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4997032394670974e-06,
+      "loss": 0.3001,
+      "step": 23949
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4991129724384224e-06,
+      "loss": 0.3336,
+      "step": 23950
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.498522812181361e-06,
+      "loss": 0.3259,
+      "step": 23951
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.497932758703331e-06,
+      "loss": 0.3192,
+      "step": 23952
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4973428120117428e-06,
+      "loss": 0.2819,
+      "step": 23953
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4967529721140029e-06,
+      "loss": 0.3332,
+      "step": 23954
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.496163239017523e-06,
+      "loss": 0.3267,
+      "step": 23955
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4955736127297082e-06,
+      "loss": 0.3234,
+      "step": 23956
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4949840932579674e-06,
+      "loss": 0.3125,
+      "step": 23957
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4943946806096998e-06,
+      "loss": 0.3333,
+      "step": 23958
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4938053747923153e-06,
+      "loss": 0.2816,
+      "step": 23959
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4932161758132101e-06,
+      "loss": 0.2899,
+      "step": 23960
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4926270836797852e-06,
+      "loss": 0.3117,
+      "step": 23961
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4920380983994398e-06,
+      "loss": 0.2856,
+      "step": 23962
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491449219979575e-06,
+      "loss": 0.3083,
+      "step": 23963
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4908604484275813e-06,
+      "loss": 0.325,
+      "step": 23964
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4902717837508585e-06,
+      "loss": 0.3025,
+      "step": 23965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4896832259567995e-06,
+      "loss": 0.3166,
+      "step": 23966
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.489094775052794e-06,
+      "loss": 0.3237,
+      "step": 23967
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.488506431046235e-06,
+      "loss": 0.2872,
+      "step": 23968
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4879181939445142e-06,
+      "loss": 0.3364,
+      "step": 23969
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.487330063755018e-06,
+      "loss": 0.3187,
+      "step": 23970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4867420404851306e-06,
+      "loss": 0.3021,
+      "step": 23971
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4861541241422417e-06,
+      "loss": 0.2881,
+      "step": 23972
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4855663147337317e-06,
+      "loss": 0.299,
+      "step": 23973
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4849786122669862e-06,
+      "loss": 0.3176,
+      "step": 23974
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.484391016749388e-06,
+      "loss": 0.2948,
+      "step": 23975
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4838035281883134e-06,
+      "loss": 0.3088,
+      "step": 23976
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4832161465911432e-06,
+      "loss": 0.3226,
+      "step": 23977
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4826288719652572e-06,
+      "loss": 0.2983,
+      "step": 23978
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4820417043180279e-06,
+      "loss": 0.3373,
+      "step": 23979
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4814546436568345e-06,
+      "loss": 0.3165,
+      "step": 23980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4808676899890472e-06,
+      "loss": 0.3062,
+      "step": 23981
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4802808433220361e-06,
+      "loss": 0.3238,
+      "step": 23982
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.479694103663175e-06,
+      "loss": 0.3304,
+      "step": 23983
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.479107471019836e-06,
+      "loss": 0.2901,
+      "step": 23984
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4785209453993821e-06,
+      "loss": 0.3118,
+      "step": 23985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4779345268091826e-06,
+      "loss": 0.314,
+      "step": 23986
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4773482152566043e-06,
+      "loss": 0.2986,
+      "step": 23987
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4767620107490078e-06,
+      "loss": 0.3263,
+      "step": 23988
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4761759132937581e-06,
+      "loss": 0.2701,
+      "step": 23989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4755899228982185e-06,
+      "loss": 0.332,
+      "step": 23990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4750040395697463e-06,
+      "loss": 0.2799,
+      "step": 23991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.474418263315699e-06,
+      "loss": 0.3192,
+      "step": 23992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4738325941434351e-06,
+      "loss": 0.3179,
+      "step": 23993
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4732470320603143e-06,
+      "loss": 0.2963,
+      "step": 23994
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4726615770736863e-06,
+      "loss": 0.3269,
+      "step": 23995
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4720762291909063e-06,
+      "loss": 0.2857,
+      "step": 23996
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4714909884193285e-06,
+      "loss": 0.3084,
+      "step": 23997
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.470905854766299e-06,
+      "loss": 0.3101,
+      "step": 23998
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.47032082823917e-06,
+      "loss": 0.296,
+      "step": 23999
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4697359088452911e-06,
+      "loss": 0.3177,
+      "step": 24000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4691510965920075e-06,
+      "loss": 0.2978,
+      "step": 24001
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.468566391486661e-06,
+      "loss": 0.3009,
+      "step": 24002
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4679817935366014e-06,
+      "loss": 0.3268,
+      "step": 24003
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4673973027491662e-06,
+      "loss": 0.2913,
+      "step": 24004
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4668129191316982e-06,
+      "loss": 0.2851,
+      "step": 24005
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4662286426915395e-06,
+      "loss": 0.2679,
+      "step": 24006
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4656444734360253e-06,
+      "loss": 0.3107,
+      "step": 24007
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4650604113724953e-06,
+      "loss": 0.2948,
+      "step": 24008
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4644764565082859e-06,
+      "loss": 0.3265,
+      "step": 24009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4638926088507277e-06,
+      "loss": 0.3494,
+      "step": 24010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4633088684071594e-06,
+      "loss": 0.3378,
+      "step": 24011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4627252351849074e-06,
+      "loss": 0.3324,
+      "step": 24012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.462141709191307e-06,
+      "loss": 0.2861,
+      "step": 24013
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4615582904336821e-06,
+      "loss": 0.3366,
+      "step": 24014
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4609749789193661e-06,
+      "loss": 0.3419,
+      "step": 24015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4603917746556796e-06,
+      "loss": 0.3211,
+      "step": 24016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4598086776499521e-06,
+      "loss": 0.3237,
+      "step": 24017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4592256879095068e-06,
+      "loss": 0.3057,
+      "step": 24018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4586428054416636e-06,
+      "loss": 0.3222,
+      "step": 24019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4580600302537452e-06,
+      "loss": 0.2992,
+      "step": 24020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4574773623530724e-06,
+      "loss": 0.324,
+      "step": 24021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.456894801746961e-06,
+      "loss": 0.2982,
+      "step": 24022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.456312348442732e-06,
+      "loss": 0.2748,
+      "step": 24023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4557300024476972e-06,
+      "loss": 0.3053,
+      "step": 24024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4551477637691713e-06,
+      "loss": 0.3339,
+      "step": 24025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4545656324144664e-06,
+      "loss": 0.3155,
+      "step": 24026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4539836083908977e-06,
+      "loss": 0.2977,
+      "step": 24027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4534016917057748e-06,
+      "loss": 0.2857,
+      "step": 24028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4528198823664019e-06,
+      "loss": 0.3015,
+      "step": 24029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.452238180380091e-06,
+      "loss": 0.2954,
+      "step": 24030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4516565857541498e-06,
+      "loss": 0.3076,
+      "step": 24031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4510750984958788e-06,
+      "loss": 0.3329,
+      "step": 24032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4504937186125844e-06,
+      "loss": 0.3134,
+      "step": 24033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4499124461115688e-06,
+      "loss": 0.245,
+      "step": 24034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4493312810001293e-06,
+      "loss": 0.3007,
+      "step": 24035
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4487502232855665e-06,
+      "loss": 0.3103,
+      "step": 24036
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.448169272975184e-06,
+      "loss": 0.3076,
+      "step": 24037
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.447588430076271e-06,
+      "loss": 0.3049,
+      "step": 24038
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4470076945961264e-06,
+      "loss": 0.282,
+      "step": 24039
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4464270665420466e-06,
+      "loss": 0.2926,
+      "step": 24040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.44584654592132e-06,
+      "loss": 0.3021,
+      "step": 24041
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4452661327412387e-06,
+      "loss": 0.3367,
+      "step": 24042
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444685827009097e-06,
+      "loss": 0.3255,
+      "step": 24043
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444105628732181e-06,
+      "loss": 0.3149,
+      "step": 24044
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4435255379177738e-06,
+      "loss": 0.32,
+      "step": 24045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4429455545731686e-06,
+      "loss": 0.2945,
+      "step": 24046
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4423656787056438e-06,
+      "loss": 0.2965,
+      "step": 24047
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4417859103224852e-06,
+      "loss": 0.318,
+      "step": 24048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4412062494309775e-06,
+      "loss": 0.2613,
+      "step": 24049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4406266960383963e-06,
+      "loss": 0.2967,
+      "step": 24050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4400472501520224e-06,
+      "loss": 0.3098,
+      "step": 24051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4394679117791376e-06,
+      "loss": 0.3316,
+      "step": 24052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.438888680927013e-06,
+      "loss": 0.3083,
+      "step": 24053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.438309557602927e-06,
+      "loss": 0.2936,
+      "step": 24054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.437730541814154e-06,
+      "loss": 0.3109,
+      "step": 24055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4371516335679625e-06,
+      "loss": 0.3319,
+      "step": 24056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4365728328716256e-06,
+      "loss": 0.3316,
+      "step": 24057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4359941397324162e-06,
+      "loss": 0.3249,
+      "step": 24058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4354155541575976e-06,
+      "loss": 0.3184,
+      "step": 24059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4348370761544383e-06,
+      "loss": 0.2915,
+      "step": 24060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4342587057302083e-06,
+      "loss": 0.2791,
+      "step": 24061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4336804428921668e-06,
+      "loss": 0.3051,
+      "step": 24062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4331022876475775e-06,
+      "loss": 0.3243,
+      "step": 24063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4325242400037065e-06,
+      "loss": 0.3304,
+      "step": 24064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4319462999678113e-06,
+      "loss": 0.3178,
+      "step": 24065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4313684675471473e-06,
+      "loss": 0.3022,
+      "step": 24066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4307907427489753e-06,
+      "loss": 0.3417,
+      "step": 24067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.430213125580554e-06,
+      "loss": 0.3078,
+      "step": 24068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.429635616049133e-06,
+      "loss": 0.3013,
+      "step": 24069
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.429058214161969e-06,
+      "loss": 0.2714,
+      "step": 24070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4284809199263172e-06,
+      "loss": 0.3222,
+      "step": 24071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4279037333494216e-06,
+      "loss": 0.2616,
+      "step": 24072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4273266544385356e-06,
+      "loss": 0.3596,
+      "step": 24073
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4267496832009088e-06,
+      "loss": 0.3317,
+      "step": 24074
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4261728196437863e-06,
+      "loss": 0.3085,
+      "step": 24075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4255960637744115e-06,
+      "loss": 0.3296,
+      "step": 24076
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.425019415600032e-06,
+      "loss": 0.2718,
+      "step": 24077
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4244428751278872e-06,
+      "loss": 0.3127,
+      "step": 24078
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4238664423652193e-06,
+      "loss": 0.3212,
+      "step": 24079
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4232901173192726e-06,
+      "loss": 0.3233,
+      "step": 24080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4227138999972801e-06,
+      "loss": 0.3241,
+      "step": 24081
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4221377904064803e-06,
+      "loss": 0.2737,
+      "step": 24082
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4215617885541132e-06,
+      "loss": 0.318,
+      "step": 24083
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4209858944474086e-06,
+      "loss": 0.3003,
+      "step": 24084
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4204101080936029e-06,
+      "loss": 0.3054,
+      "step": 24085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4198344294999277e-06,
+      "loss": 0.2952,
+      "step": 24086
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.41925885867361e-06,
+      "loss": 0.3116,
+      "step": 24087
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4186833956218815e-06,
+      "loss": 0.3357,
+      "step": 24088
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.418108040351972e-06,
+      "loss": 0.3489,
+      "step": 24089
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4175327928711047e-06,
+      "loss": 0.3367,
+      "step": 24090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4169576531865058e-06,
+      "loss": 0.2876,
+      "step": 24091
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4163826213054022e-06,
+      "loss": 0.315,
+      "step": 24092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.415807697235011e-06,
+      "loss": 0.3394,
+      "step": 24093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4152328809825555e-06,
+      "loss": 0.3037,
+      "step": 24094
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4146581725552588e-06,
+      "loss": 0.3017,
+      "step": 24095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4140835719603363e-06,
+      "loss": 0.2794,
+      "step": 24096
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4135090792050022e-06,
+      "loss": 0.2858,
+      "step": 24097
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4129346942964773e-06,
+      "loss": 0.2811,
+      "step": 24098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4123604172419714e-06,
+      "loss": 0.3535,
+      "step": 24099
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4117862480486998e-06,
+      "loss": 0.3137,
+      "step": 24100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4112121867238738e-06,
+      "loss": 0.3392,
+      "step": 24101
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4106382332747059e-06,
+      "loss": 0.3393,
+      "step": 24102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.410064387708401e-06,
+      "loss": 0.3186,
+      "step": 24103
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.409490650032167e-06,
+      "loss": 0.3375,
+      "step": 24104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4089170202532155e-06,
+      "loss": 0.3123,
+      "step": 24105
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.408343498378747e-06,
+      "loss": 0.3041,
+      "step": 24106
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4077700844159624e-06,
+      "loss": 0.3243,
+      "step": 24107
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4071967783720696e-06,
+      "loss": 0.3278,
+      "step": 24108
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4066235802542649e-06,
+      "loss": 0.2891,
+      "step": 24109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.406050490069748e-06,
+      "loss": 0.3234,
+      "step": 24110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.405477507825721e-06,
+      "loss": 0.2981,
+      "step": 24111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.404904633529376e-06,
+      "loss": 0.3146,
+      "step": 24112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4043318671879102e-06,
+      "loss": 0.2818,
+      "step": 24113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4037592088085195e-06,
+      "loss": 0.2643,
+      "step": 24114
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4031866583983923e-06,
+      "loss": 0.3176,
+      "step": 24115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4026142159647239e-06,
+      "loss": 0.3456,
+      "step": 24116
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4020418815147007e-06,
+      "loss": 0.3966,
+      "step": 24117
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4014696550555152e-06,
+      "loss": 0.3269,
+      "step": 24118
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4008975365943511e-06,
+      "loss": 0.3153,
+      "step": 24119
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4003255261383963e-06,
+      "loss": 0.2622,
+      "step": 24120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3997536236948327e-06,
+      "loss": 0.2622,
+      "step": 24121
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.399181829270846e-06,
+      "loss": 0.2919,
+      "step": 24122
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3986101428736188e-06,
+      "loss": 0.3325,
+      "step": 24123
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3980385645103277e-06,
+      "loss": 0.2919,
+      "step": 24124
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.397467094188154e-06,
+      "loss": 0.3172,
+      "step": 24125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3968957319142785e-06,
+      "loss": 0.3019,
+      "step": 24126
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3963244776958706e-06,
+      "loss": 0.3062,
+      "step": 24127
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3957533315401127e-06,
+      "loss": 0.3147,
+      "step": 24128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3951822934541749e-06,
+      "loss": 0.3211,
+      "step": 24129
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3946113634452263e-06,
+      "loss": 0.3195,
+      "step": 24130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3940405415204417e-06,
+      "loss": 0.2765,
+      "step": 24131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3934698276869907e-06,
+      "loss": 0.3136,
+      "step": 24132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.39289922195204e-06,
+      "loss": 0.3274,
+      "step": 24133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3923287243227568e-06,
+      "loss": 0.3452,
+      "step": 24134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.391758334806309e-06,
+      "loss": 0.2969,
+      "step": 24135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3911880534098565e-06,
+      "loss": 0.3217,
+      "step": 24136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3906178801405634e-06,
+      "loss": 0.3369,
+      "step": 24137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3900478150055952e-06,
+      "loss": 0.3226,
+      "step": 24138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3894778580121083e-06,
+      "loss": 0.3248,
+      "step": 24139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.388908009167259e-06,
+      "loss": 0.3071,
+      "step": 24140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3883382684782077e-06,
+      "loss": 0.3111,
+      "step": 24141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3877686359521126e-06,
+      "loss": 0.3075,
+      "step": 24142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3871991115961237e-06,
+      "loss": 0.2945,
+      "step": 24143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3866296954173952e-06,
+      "loss": 0.3238,
+      "step": 24144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3860603874230838e-06,
+      "loss": 0.2568,
+      "step": 24145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3854911876203324e-06,
+      "loss": 0.3078,
+      "step": 24146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3849220960162956e-06,
+      "loss": 0.3317,
+      "step": 24147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.384353112618122e-06,
+      "loss": 0.3181,
+      "step": 24148
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3837842374329557e-06,
+      "loss": 0.3116,
+      "step": 24149
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.383215470467939e-06,
+      "loss": 0.306,
+      "step": 24150
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3826468117302216e-06,
+      "loss": 0.3171,
+      "step": 24151
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3820782612269401e-06,
+      "loss": 0.3564,
+      "step": 24152
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3815098189652388e-06,
+      "loss": 0.2963,
+      "step": 24153
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3809414849522585e-06,
+      "loss": 0.3208,
+      "step": 24154
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3803732591951336e-06,
+      "loss": 0.2828,
+      "step": 24155
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3798051417010039e-06,
+      "loss": 0.3516,
+      "step": 24156
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.379237132477006e-06,
+      "loss": 0.2853,
+      "step": 24157
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3786692315302707e-06,
+      "loss": 0.28,
+      "step": 24158
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3781014388679349e-06,
+      "loss": 0.2762,
+      "step": 24159
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3775337544971279e-06,
+      "loss": 0.3167,
+      "step": 24160
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3769661784249776e-06,
+      "loss": 0.2921,
+      "step": 24161
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3763987106586152e-06,
+      "loss": 0.3443,
+      "step": 24162
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3758313512051702e-06,
+      "loss": 0.3107,
+      "step": 24163
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3752641000717649e-06,
+      "loss": 0.3033,
+      "step": 24164
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3746969572655245e-06,
+      "loss": 0.3319,
+      "step": 24165
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3741299227935757e-06,
+      "loss": 0.3333,
+      "step": 24166
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3735629966630359e-06,
+      "loss": 0.3304,
+      "step": 24167
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3729961788810288e-06,
+      "loss": 0.2903,
+      "step": 24168
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3724294694546736e-06,
+      "loss": 0.3367,
+      "step": 24169
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3718628683910884e-06,
+      "loss": 0.3014,
+      "step": 24170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3712963756973863e-06,
+      "loss": 0.3316,
+      "step": 24171
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3707299913806883e-06,
+      "loss": 0.3348,
+      "step": 24172
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3701637154481018e-06,
+      "loss": 0.3048,
+      "step": 24173
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3695975479067425e-06,
+      "loss": 0.2939,
+      "step": 24174
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.36903148876372e-06,
+      "loss": 0.2681,
+      "step": 24175
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3684655380261492e-06,
+      "loss": 0.3227,
+      "step": 24176
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3678996957011325e-06,
+      "loss": 0.3178,
+      "step": 24177
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3673339617957782e-06,
+      "loss": 0.3087,
+      "step": 24178
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3667683363171957e-06,
+      "loss": 0.3045,
+      "step": 24179
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3662028192724864e-06,
+      "loss": 0.3062,
+      "step": 24180
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3656374106687508e-06,
+      "loss": 0.3359,
+      "step": 24181
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3650721105130948e-06,
+      "loss": 0.2993,
+      "step": 24182
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3645069188126147e-06,
+      "loss": 0.3009,
+      "step": 24183
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3639418355744116e-06,
+      "loss": 0.2995,
+      "step": 24184
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3633768608055854e-06,
+      "loss": 0.3041,
+      "step": 24185
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3628119945132268e-06,
+      "loss": 0.2725,
+      "step": 24186
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3622472367044325e-06,
+      "loss": 0.3428,
+      "step": 24187
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3616825873862994e-06,
+      "loss": 0.284,
+      "step": 24188
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3611180465659156e-06,
+      "loss": 0.3196,
+      "step": 24189
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3605536142503739e-06,
+      "loss": 0.3335,
+      "step": 24190
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3599892904467637e-06,
+      "loss": 0.3022,
+      "step": 24191
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3594250751621686e-06,
+      "loss": 0.3294,
+      "step": 24192
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.358860968403679e-06,
+      "loss": 0.2919,
+      "step": 24193
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3582969701783822e-06,
+      "loss": 0.3289,
+      "step": 24194
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3577330804933563e-06,
+      "loss": 0.2991,
+      "step": 24195
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3571692993556874e-06,
+      "loss": 0.3308,
+      "step": 24196
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3566056267724593e-06,
+      "loss": 0.3206,
+      "step": 24197
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3560420627507453e-06,
+      "loss": 0.3177,
+      "step": 24198
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3554786072976266e-06,
+      "loss": 0.3083,
+      "step": 24199
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3549152604201843e-06,
+      "loss": 0.2545,
+      "step": 24200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3543520221254902e-06,
+      "loss": 0.3029,
+      "step": 24201
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3537888924206156e-06,
+      "loss": 0.3306,
+      "step": 24202
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.35322587131264e-06,
+      "loss": 0.3405,
+      "step": 24203
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3526629588086294e-06,
+      "loss": 0.3191,
+      "step": 24204
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3521001549156565e-06,
+      "loss": 0.3037,
+      "step": 24205
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3515374596407916e-06,
+      "loss": 0.3425,
+      "step": 24206
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3509748729910988e-06,
+      "loss": 0.312,
+      "step": 24207
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3504123949736458e-06,
+      "loss": 0.314,
+      "step": 24208
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349850025595497e-06,
+      "loss": 0.302,
+      "step": 24209
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3492877648637204e-06,
+      "loss": 0.3163,
+      "step": 24210
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.348725612785372e-06,
+      "loss": 0.3093,
+      "step": 24211
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3481635693675131e-06,
+      "loss": 0.3038,
+      "step": 24212
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.347601634617206e-06,
+      "loss": 0.2848,
+      "step": 24213
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.347039808541506e-06,
+      "loss": 0.3218,
+      "step": 24214
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3464780911474695e-06,
+      "loss": 0.3161,
+      "step": 24215
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3459164824421556e-06,
+      "loss": 0.318,
+      "step": 24216
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3453549824326128e-06,
+      "loss": 0.2725,
+      "step": 24217
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3447935911258957e-06,
+      "loss": 0.306,
+      "step": 24218
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3442323085290588e-06,
+      "loss": 0.3123,
+      "step": 24219
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.343671134649146e-06,
+      "loss": 0.287,
+      "step": 24220
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.343110069493212e-06,
+      "loss": 0.3279,
+      "step": 24221
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3425491130682965e-06,
+      "loss": 0.2607,
+      "step": 24222
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3419882653814532e-06,
+      "loss": 0.2993,
+      "step": 24223
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341427526439718e-06,
+      "loss": 0.2733,
+      "step": 24224
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3408668962501415e-06,
+      "loss": 0.3135,
+      "step": 24225
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3403063748197586e-06,
+      "loss": 0.3199,
+      "step": 24226
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.3033,
+      "step": 24227
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3391856582647455e-06,
+      "loss": 0.3077,
+      "step": 24228
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3386254631541896e-06,
+      "loss": 0.2621,
+      "step": 24229
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3380653768309827e-06,
+      "loss": 0.2923,
+      "step": 24230
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3375053993021613e-06,
+      "loss": 0.309,
+      "step": 24231
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3369455305747547e-06,
+      "loss": 0.3748,
+      "step": 24232
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3363857706558003e-06,
+      "loss": 0.3046,
+      "step": 24233
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3358261195523258e-06,
+      "loss": 0.3557,
+      "step": 24234
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.335266577271359e-06,
+      "loss": 0.3287,
+      "step": 24235
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3347071438199288e-06,
+      "loss": 0.3286,
+      "step": 24236
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.334147819205064e-06,
+      "loss": 0.3019,
+      "step": 24237
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3335886034337854e-06,
+      "loss": 0.3226,
+      "step": 24238
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3330294965131197e-06,
+      "loss": 0.2797,
+      "step": 24239
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3324704984500914e-06,
+      "loss": 0.2717,
+      "step": 24240
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.331911609251717e-06,
+      "loss": 0.2769,
+      "step": 24241
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3313528289250176e-06,
+      "loss": 0.296,
+      "step": 24242
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3307941574770156e-06,
+      "loss": 0.3461,
+      "step": 24243
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3302355949147227e-06,
+      "loss": 0.2913,
+      "step": 24244
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3296771412451558e-06,
+      "loss": 0.3138,
+      "step": 24245
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3291187964753283e-06,
+      "loss": 0.2905,
+      "step": 24246
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3285605606122576e-06,
+      "loss": 0.314,
+      "step": 24247
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3280024336629493e-06,
+      "loss": 0.2969,
+      "step": 24248
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3274444156344157e-06,
+      "loss": 0.262,
+      "step": 24249
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.326886506533669e-06,
+      "loss": 0.3103,
+      "step": 24250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.32632870636771e-06,
+      "loss": 0.3514,
+      "step": 24251
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3257710151435489e-06,
+      "loss": 0.309,
+      "step": 24252
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.325213432868192e-06,
+      "loss": 0.2871,
+      "step": 24253
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3246559595486408e-06,
+      "loss": 0.2918,
+      "step": 24254
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3240985951918939e-06,
+      "loss": 0.3226,
+      "step": 24255
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3235413398049568e-06,
+      "loss": 0.3179,
+      "step": 24256
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3229841933948239e-06,
+      "loss": 0.2675,
+      "step": 24257
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3224271559684953e-06,
+      "loss": 0.3377,
+      "step": 24258
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3218702275329708e-06,
+      "loss": 0.2851,
+      "step": 24259
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3213134080952394e-06,
+      "loss": 0.2998,
+      "step": 24260
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3207566976622977e-06,
+      "loss": 0.3017,
+      "step": 24261
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.32020009624114e-06,
+      "loss": 0.3081,
+      "step": 24262
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3196436038387527e-06,
+      "loss": 0.3122,
+      "step": 24263
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3190872204621308e-06,
+      "loss": 0.3089,
+      "step": 24264
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3185309461182593e-06,
+      "loss": 0.2914,
+      "step": 24265
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3179747808141241e-06,
+      "loss": 0.3154,
+      "step": 24266
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3174187245567105e-06,
+      "loss": 0.3433,
+      "step": 24267
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3168627773530073e-06,
+      "loss": 0.3143,
+      "step": 24268
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3163069392099903e-06,
+      "loss": 0.2917,
+      "step": 24269
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3157512101346458e-06,
+      "loss": 0.2963,
+      "step": 24270
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3151955901339543e-06,
+      "loss": 0.3098,
+      "step": 24271
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3146400792148907e-06,
+      "loss": 0.2851,
+      "step": 24272
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.314084677384433e-06,
+      "loss": 0.3133,
+      "step": 24273
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.313529384649561e-06,
+      "loss": 0.2746,
+      "step": 24274
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3129742010172476e-06,
+      "loss": 0.3148,
+      "step": 24275
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3124191264944608e-06,
+      "loss": 0.3198,
+      "step": 24276
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.31186416108818e-06,
+      "loss": 0.2739,
+      "step": 24277
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3113093048053682e-06,
+      "loss": 0.3124,
+      "step": 24278
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.310754557652999e-06,
+      "loss": 0.2987,
+      "step": 24279
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3101999196380422e-06,
+      "loss": 0.2992,
+      "step": 24280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3096453907674578e-06,
+      "loss": 0.3036,
+      "step": 24281
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3090909710482136e-06,
+      "loss": 0.2812,
+      "step": 24282
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.308536660487275e-06,
+      "loss": 0.3656,
+      "step": 24283
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3079824590916035e-06,
+      "loss": 0.2913,
+      "step": 24284
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3074283668681598e-06,
+      "loss": 0.324,
+      "step": 24285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3068743838239005e-06,
+      "loss": 0.3137,
+      "step": 24286
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3063205099657883e-06,
+      "loss": 0.2853,
+      "step": 24287
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3057667453007749e-06,
+      "loss": 0.2957,
+      "step": 24288
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3052130898358173e-06,
+      "loss": 0.3035,
+      "step": 24289
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3046595435778741e-06,
+      "loss": 0.3107,
+      "step": 24290
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.30410610653389e-06,
+      "loss": 0.3218,
+      "step": 24291
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3035527787108215e-06,
+      "loss": 0.3148,
+      "step": 24292
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3029995601156187e-06,
+      "loss": 0.3224,
+      "step": 24293
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.302446450755227e-06,
+      "loss": 0.314,
+      "step": 24294
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3018934506365967e-06,
+      "loss": 0.3202,
+      "step": 24295
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3013405597666694e-06,
+      "loss": 0.333,
+      "step": 24296
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3007877781523948e-06,
+      "loss": 0.2896,
+      "step": 24297
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.30023510580071e-06,
+      "loss": 0.2933,
+      "step": 24298
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2996825427185622e-06,
+      "loss": 0.2866,
+      "step": 24299
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2991300889128867e-06,
+      "loss": 0.2813,
+      "step": 24300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2985777443906244e-06,
+      "loss": 0.2821,
+      "step": 24301
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2980255091587147e-06,
+      "loss": 0.2892,
+      "step": 24302
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2974733832240904e-06,
+      "loss": 0.2957,
+      "step": 24303
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2969213665936875e-06,
+      "loss": 0.3397,
+      "step": 24304
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2963694592744425e-06,
+      "loss": 0.3319,
+      "step": 24305
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.295817661273282e-06,
+      "loss": 0.2709,
+      "step": 24306
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2952659725971406e-06,
+      "loss": 0.2852,
+      "step": 24307
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2947143932529472e-06,
+      "loss": 0.2994,
+      "step": 24308
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.294162923247626e-06,
+      "loss": 0.3051,
+      "step": 24309
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2936115625881051e-06,
+      "loss": 0.319,
+      "step": 24310
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2930603112813145e-06,
+      "loss": 0.3058,
+      "step": 24311
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2925091693341706e-06,
+      "loss": 0.3059,
+      "step": 24312
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2919581367535993e-06,
+      "loss": 0.2798,
+      "step": 24313
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.291407213546524e-06,
+      "loss": 0.2894,
+      "step": 24314
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2908563997198598e-06,
+      "loss": 0.3028,
+      "step": 24315
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2903056952805271e-06,
+      "loss": 0.3146,
+      "step": 24316
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2897551002354448e-06,
+      "loss": 0.3253,
+      "step": 24317
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.289204614591526e-06,
+      "loss": 0.3099,
+      "step": 24318
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2886542383556832e-06,
+      "loss": 0.3243,
+      "step": 24319
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2881039715348298e-06,
+      "loss": 0.3254,
+      "step": 24320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2875538141358822e-06,
+      "loss": 0.2911,
+      "step": 24321
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2870037661657443e-06,
+      "loss": 0.2991,
+      "step": 24322
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2864538276313265e-06,
+      "loss": 0.3045,
+      "step": 24323
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2859039985395383e-06,
+      "loss": 0.3023,
+      "step": 24324
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2853542788972827e-06,
+      "loss": 0.3256,
+      "step": 24325
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2848046687114678e-06,
+      "loss": 0.3374,
+      "step": 24326
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2842551679889915e-06,
+      "loss": 0.3228,
+      "step": 24327
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2837057767367612e-06,
+      "loss": 0.2754,
+      "step": 24328
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2831564949616714e-06,
+      "loss": 0.3039,
+      "step": 24329
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.282607322670627e-06,
+      "loss": 0.305,
+      "step": 24330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.282058259870521e-06,
+      "loss": 0.3256,
+      "step": 24331
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2815093065682505e-06,
+      "loss": 0.3067,
+      "step": 24332
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.280960462770715e-06,
+      "loss": 0.268,
+      "step": 24333
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2804117284848017e-06,
+      "loss": 0.3037,
+      "step": 24334
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2798631037174048e-06,
+      "loss": 0.318,
+      "step": 24335
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2793145884754177e-06,
+      "loss": 0.2881,
+      "step": 24336
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2787661827657272e-06,
+      "loss": 0.3122,
+      "step": 24337
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.278217886595222e-06,
+      "loss": 0.3016,
+      "step": 24338
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2776696999707905e-06,
+      "loss": 0.2589,
+      "step": 24339
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2771216228993132e-06,
+      "loss": 0.3108,
+      "step": 24340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2765736553876762e-06,
+      "loss": 0.2939,
+      "step": 24341
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.276025797442766e-06,
+      "loss": 0.317,
+      "step": 24342
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.275478049071457e-06,
+      "loss": 0.3355,
+      "step": 24343
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2749304102806327e-06,
+      "loss": 0.3334,
+      "step": 24344
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.274382881077174e-06,
+      "loss": 0.316,
+      "step": 24345
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2738354614679516e-06,
+      "loss": 0.3187,
+      "step": 24346
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.273288151459846e-06,
+      "loss": 0.3205,
+      "step": 24347
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.272740951059731e-06,
+      "loss": 0.3185,
+      "step": 24348
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2721938602744787e-06,
+      "loss": 0.2833,
+      "step": 24349
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2716468791109583e-06,
+      "loss": 0.3074,
+      "step": 24350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2711000075760437e-06,
+      "loss": 0.292,
+      "step": 24351
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2705532456766e-06,
+      "loss": 0.3099,
+      "step": 24352
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.270006593419497e-06,
+      "loss": 0.2894,
+      "step": 24353
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2694600508116018e-06,
+      "loss": 0.2901,
+      "step": 24354
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.268913617859775e-06,
+      "loss": 0.3039,
+      "step": 24355
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2683672945708826e-06,
+      "loss": 0.3385,
+      "step": 24356
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2678210809517855e-06,
+      "loss": 0.2923,
+      "step": 24357
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2672749770093473e-06,
+      "loss": 0.2885,
+      "step": 24358
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2667289827504237e-06,
+      "loss": 0.2541,
+      "step": 24359
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2661830981818723e-06,
+      "loss": 0.2956,
+      "step": 24360
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2656373233105513e-06,
+      "loss": 0.3144,
+      "step": 24361
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2650916581433137e-06,
+      "loss": 0.3067,
+      "step": 24362
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2645461026870133e-06,
+      "loss": 0.2971,
+      "step": 24363
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2640006569485053e-06,
+      "loss": 0.309,
+      "step": 24364
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2634553209346356e-06,
+      "loss": 0.2788,
+      "step": 24365
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2629100946522577e-06,
+      "loss": 0.3118,
+      "step": 24366
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2623649781082192e-06,
+      "loss": 0.3033,
+      "step": 24367
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2618199713093649e-06,
+      "loss": 0.2728,
+      "step": 24368
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2612750742625424e-06,
+      "loss": 0.2706,
+      "step": 24369
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2607302869745953e-06,
+      "loss": 0.2889,
+      "step": 24370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2601856094523624e-06,
+      "loss": 0.3156,
+      "step": 24371
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2596410417026872e-06,
+      "loss": 0.313,
+      "step": 24372
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2590965837324132e-06,
+      "loss": 0.3144,
+      "step": 24373
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2585522355483725e-06,
+      "loss": 0.3157,
+      "step": 24374
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2580079971574066e-06,
+      "loss": 0.3001,
+      "step": 24375
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.257463868566351e-06,
+      "loss": 0.345,
+      "step": 24376
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2569198497820379e-06,
+      "loss": 0.3148,
+      "step": 24377
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2563759408112997e-06,
+      "loss": 0.3087,
+      "step": 24378
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.255832141660972e-06,
+      "loss": 0.3095,
+      "step": 24379
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2552884523378828e-06,
+      "loss": 0.306,
+      "step": 24380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2547448728488576e-06,
+      "loss": 0.3284,
+      "step": 24381
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2542014032007287e-06,
+      "loss": 0.3032,
+      "step": 24382
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2536580434003187e-06,
+      "loss": 0.3063,
+      "step": 24383
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.253114793454453e-06,
+      "loss": 0.3191,
+      "step": 24384
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2525716533699572e-06,
+      "loss": 0.3251,
+      "step": 24385
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2520286231536505e-06,
+      "loss": 0.3043,
+      "step": 24386
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.251485702812353e-06,
+      "loss": 0.3055,
+      "step": 24387
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2509428923528876e-06,
+      "loss": 0.3005,
+      "step": 24388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2504001917820673e-06,
+      "loss": 0.2928,
+      "step": 24389
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249857601106712e-06,
+      "loss": 0.2679,
+      "step": 24390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249315120333634e-06,
+      "loss": 0.3219,
+      "step": 24391
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.24877274946965e-06,
+      "loss": 0.2891,
+      "step": 24392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.248230488521569e-06,
+      "loss": 0.3038,
+      "step": 24393
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2476883374962024e-06,
+      "loss": 0.3079,
+      "step": 24394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2471462964003634e-06,
+      "loss": 0.3046,
+      "step": 24395
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2466043652408533e-06,
+      "loss": 0.3154,
+      "step": 24396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2460625440244844e-06,
+      "loss": 0.2904,
+      "step": 24397
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2455208327580614e-06,
+      "loss": 0.3208,
+      "step": 24398
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2449792314483855e-06,
+      "loss": 0.2911,
+      "step": 24399
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2444377401022623e-06,
+      "loss": 0.3092,
+      "step": 24400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2438963587264896e-06,
+      "loss": 0.3003,
+      "step": 24401
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.243355087327871e-06,
+      "loss": 0.3064,
+      "step": 24402
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.242813925913201e-06,
+      "loss": 0.3244,
+      "step": 24403
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.242272874489281e-06,
+      "loss": 0.3007,
+      "step": 24404
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.241731933062902e-06,
+      "loss": 0.3199,
+      "step": 24405
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2411911016408595e-06,
+      "loss": 0.3243,
+      "step": 24406
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2406503802299508e-06,
+      "loss": 0.3044,
+      "step": 24407
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2401097688369612e-06,
+      "loss": 0.3116,
+      "step": 24408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2395692674686832e-06,
+      "loss": 0.3331,
+      "step": 24409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.239028876131908e-06,
+      "loss": 0.2571,
+      "step": 24410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.238488594833419e-06,
+      "loss": 0.3072,
+      "step": 24411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2379484235800054e-06,
+      "loss": 0.2831,
+      "step": 24412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2374083623784516e-06,
+      "loss": 0.3053,
+      "step": 24413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2368684112355366e-06,
+      "loss": 0.309,
+      "step": 24414
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2363285701580451e-06,
+      "loss": 0.3177,
+      "step": 24415
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2357888391527596e-06,
+      "loss": 0.3099,
+      "step": 24416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2352492182264552e-06,
+      "loss": 0.3091,
+      "step": 24417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2347097073859116e-06,
+      "loss": 0.3061,
+      "step": 24418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2341703066379073e-06,
+      "loss": 0.3002,
+      "step": 24419
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2336310159892118e-06,
+      "loss": 0.3021,
+      "step": 24420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2330918354466026e-06,
+      "loss": 0.3077,
+      "step": 24421
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2325527650168533e-06,
+      "loss": 0.3175,
+      "step": 24422
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.232013804706731e-06,
+      "loss": 0.3084,
+      "step": 24423
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2314749545230053e-06,
+      "loss": 0.3056,
+      "step": 24424
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2309362144724479e-06,
+      "loss": 0.3123,
+      "step": 24425
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2303975845618198e-06,
+      "loss": 0.2693,
+      "step": 24426
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2298590647978892e-06,
+      "loss": 0.2894,
+      "step": 24427
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2293206551874193e-06,
+      "loss": 0.2873,
+      "step": 24428
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.228782355737177e-06,
+      "loss": 0.359,
+      "step": 24429
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2282441664539157e-06,
+      "loss": 0.32,
+      "step": 24430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2277060873444003e-06,
+      "loss": 0.3205,
+      "step": 24431
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2271681184153895e-06,
+      "loss": 0.3033,
+      "step": 24432
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.226630259673638e-06,
+      "loss": 0.3134,
+      "step": 24433
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2260925111259003e-06,
+      "loss": 0.3409,
+      "step": 24434
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2255548727789336e-06,
+      "loss": 0.3328,
+      "step": 24435
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2250173446394876e-06,
+      "loss": 0.3198,
+      "step": 24436
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2244799267143138e-06,
+      "loss": 0.3125,
+      "step": 24437
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2239426190101667e-06,
+      "loss": 0.3045,
+      "step": 24438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2234054215337899e-06,
+      "loss": 0.2841,
+      "step": 24439
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2228683342919323e-06,
+      "loss": 0.3046,
+      "step": 24440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2223313572913432e-06,
+      "loss": 0.3473,
+      "step": 24441
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2217944905387602e-06,
+      "loss": 0.3059,
+      "step": 24442
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2212577340409327e-06,
+      "loss": 0.2979,
+      "step": 24443
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2207210878045993e-06,
+      "loss": 0.3157,
+      "step": 24444
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2201845518364996e-06,
+      "loss": 0.2904,
+      "step": 24445
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2196481261433735e-06,
+      "loss": 0.3323,
+      "step": 24446
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2191118107319599e-06,
+      "loss": 0.3224,
+      "step": 24447
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2185756056089925e-06,
+      "loss": 0.3079,
+      "step": 24448
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.218039510781207e-06,
+      "loss": 0.2977,
+      "step": 24449
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2175035262553392e-06,
+      "loss": 0.2883,
+      "step": 24450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2169676520381168e-06,
+      "loss": 0.3369,
+      "step": 24451
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2164318881362725e-06,
+      "loss": 0.3391,
+      "step": 24452
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2158962345565373e-06,
+      "loss": 0.2945,
+      "step": 24453
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2153606913056381e-06,
+      "loss": 0.3155,
+      "step": 24454
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2148252583902987e-06,
+      "loss": 0.3163,
+      "step": 24455
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2142899358172476e-06,
+      "loss": 0.2967,
+      "step": 24456
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2137547235932046e-06,
+      "loss": 0.3277,
+      "step": 24457
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2132196217248948e-06,
+      "loss": 0.3538,
+      "step": 24458
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.212684630219041e-06,
+      "loss": 0.3212,
+      "step": 24459
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2121497490823587e-06,
+      "loss": 0.2805,
+      "step": 24460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2116149783215668e-06,
+      "loss": 0.3338,
+      "step": 24461
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.211080317943385e-06,
+      "loss": 0.3027,
+      "step": 24462
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2105457679545252e-06,
+      "loss": 0.3237,
+      "step": 24463
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2100113283617043e-06,
+      "loss": 0.3137,
+      "step": 24464
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2094769991716314e-06,
+      "loss": 0.295,
+      "step": 24465
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2089427803910225e-06,
+      "loss": 0.286,
+      "step": 24466
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2084086720265819e-06,
+      "loss": 0.3258,
+      "step": 24467
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.207874674085021e-06,
+      "loss": 0.2941,
+      "step": 24468
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2073407865730491e-06,
+      "loss": 0.3271,
+      "step": 24469
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206807009497366e-06,
+      "loss": 0.305,
+      "step": 24470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.20627334286468e-06,
+      "loss": 0.2529,
+      "step": 24471
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2057397866816956e-06,
+      "loss": 0.3011,
+      "step": 24472
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2052063409551108e-06,
+      "loss": 0.3379,
+      "step": 24473
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.204673005691628e-06,
+      "loss": 0.3226,
+      "step": 24474
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2041397808979448e-06,
+      "loss": 0.3268,
+      "step": 24475
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2036066665807565e-06,
+      "loss": 0.3172,
+      "step": 24476
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203073662746761e-06,
+      "loss": 0.3274,
+      "step": 24477
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2025407694026559e-06,
+      "loss": 0.2924,
+      "step": 24478
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.202007986555128e-06,
+      "loss": 0.2829,
+      "step": 24479
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2014753142108738e-06,
+      "loss": 0.312,
+      "step": 24480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2009427523765837e-06,
+      "loss": 0.32,
+      "step": 24481
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2004103010589441e-06,
+      "loss": 0.3133,
+      "step": 24482
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1998779602646438e-06,
+      "loss": 0.3175,
+      "step": 24483
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1993457300003708e-06,
+      "loss": 0.3104,
+      "step": 24484
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1988136102728076e-06,
+      "loss": 0.3143,
+      "step": 24485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1982816010886379e-06,
+      "loss": 0.3096,
+      "step": 24486
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.197749702454546e-06,
+      "loss": 0.2891,
+      "step": 24487
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1972179143772078e-06,
+      "loss": 0.3241,
+      "step": 24488
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1966862368633059e-06,
+      "loss": 0.2755,
+      "step": 24489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1961546699195204e-06,
+      "loss": 0.3039,
+      "step": 24490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1956232135525237e-06,
+      "loss": 0.3095,
+      "step": 24491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1950918677689916e-06,
+      "loss": 0.3339,
+      "step": 24492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.194560632575601e-06,
+      "loss": 0.2781,
+      "step": 24493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1940295079790199e-06,
+      "loss": 0.3403,
+      "step": 24494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.193498493985924e-06,
+      "loss": 0.3135,
+      "step": 24495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.192967590602978e-06,
+      "loss": 0.3202,
+      "step": 24496
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1924367978368535e-06,
+      "loss": 0.3239,
+      "step": 24497
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1919061156942147e-06,
+      "loss": 0.2849,
+      "step": 24498
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1913755441817309e-06,
+      "loss": 0.3167,
+      "step": 24499
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1908450833060614e-06,
+      "loss": 0.312,
+      "step": 24500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1903147330738706e-06,
+      "loss": 0.3211,
+      "step": 24501
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1897844934918201e-06,
+      "loss": 0.326,
+      "step": 24502
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1892543645665721e-06,
+      "loss": 0.3116,
+      "step": 24503
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1887243463047804e-06,
+      "loss": 0.3282,
+      "step": 24504
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1881944387131071e-06,
+      "loss": 0.3204,
+      "step": 24505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.187664641798203e-06,
+      "loss": 0.2821,
+      "step": 24506
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1871349555667255e-06,
+      "loss": 0.3114,
+      "step": 24507
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1866053800253253e-06,
+      "loss": 0.3298,
+      "step": 24508
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1860759151806579e-06,
+      "loss": 0.3342,
+      "step": 24509
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1855465610393669e-06,
+      "loss": 0.2764,
+      "step": 24510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1850173176081058e-06,
+      "loss": 0.2801,
+      "step": 24511
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.184488184893523e-06,
+      "loss": 0.3084,
+      "step": 24512
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1839591629022607e-06,
+      "loss": 0.2861,
+      "step": 24513
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1834302516409635e-06,
+      "loss": 0.3306,
+      "step": 24514
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1829014511162794e-06,
+      "loss": 0.2773,
+      "step": 24515
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1823727613348445e-06,
+      "loss": 0.3103,
+      "step": 24516
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1818441823033045e-06,
+      "loss": 0.3417,
+      "step": 24517
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1813157140282937e-06,
+      "loss": 0.3059,
+      "step": 24518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1807873565164507e-06,
+      "loss": 0.3062,
+      "step": 24519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1802591097744121e-06,
+      "loss": 0.2817,
+      "step": 24520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1797309738088147e-06,
+      "loss": 0.3011,
+      "step": 24521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.179202948626289e-06,
+      "loss": 0.3304,
+      "step": 24522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1786750342334674e-06,
+      "loss": 0.3156,
+      "step": 24523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1781472306369845e-06,
+      "loss": 0.3411,
+      "step": 24524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.177619537843464e-06,
+      "loss": 0.3216,
+      "step": 24525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.177091955859535e-06,
+      "loss": 0.3037,
+      "step": 24526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.176564484691829e-06,
+      "loss": 0.2614,
+      "step": 24527
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.176037124346966e-06,
+      "loss": 0.3132,
+      "step": 24528
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1755098748315696e-06,
+      "loss": 0.3069,
+      "step": 24529
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1749827361522658e-06,
+      "loss": 0.3243,
+      "step": 24530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1744557083156705e-06,
+      "loss": 0.313,
+      "step": 24531
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1739287913284047e-06,
+      "loss": 0.3057,
+      "step": 24532
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1734019851970912e-06,
+      "loss": 0.3017,
+      "step": 24533
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1728752899283401e-06,
+      "loss": 0.3174,
+      "step": 24534
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1723487055287709e-06,
+      "loss": 0.2934,
+      "step": 24535
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1718222320049976e-06,
+      "loss": 0.3415,
+      "step": 24536
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.171295869363629e-06,
+      "loss": 0.3408,
+      "step": 24537
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1707696176112815e-06,
+      "loss": 0.3331,
+      "step": 24538
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.17024347675456e-06,
+      "loss": 0.268,
+      "step": 24539
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1697174468000772e-06,
+      "loss": 0.2973,
+      "step": 24540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1691915277544351e-06,
+      "loss": 0.301,
+      "step": 24541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1686657196242423e-06,
+      "loss": 0.3121,
+      "step": 24542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1681400224161043e-06,
+      "loss": 0.3004,
+      "step": 24543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1676144361366214e-06,
+      "loss": 0.2894,
+      "step": 24544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1670889607923953e-06,
+      "loss": 0.3101,
+      "step": 24545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1665635963900278e-06,
+      "loss": 0.357,
+      "step": 24546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1660383429361155e-06,
+      "loss": 0.294,
+      "step": 24547
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1655132004372572e-06,
+      "loss": 0.3095,
+      "step": 24548
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1649881689000487e-06,
+      "loss": 0.2955,
+      "step": 24549
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1644632483310814e-06,
+      "loss": 0.2802,
+      "step": 24550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.16393843873695e-06,
+      "loss": 0.2798,
+      "step": 24551
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1634137401242496e-06,
+      "loss": 0.3183,
+      "step": 24552
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1628891524995645e-06,
+      "loss": 0.3279,
+      "step": 24553
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1623646758694873e-06,
+      "loss": 0.2913,
+      "step": 24554
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1618403102406073e-06,
+      "loss": 0.3329,
+      "step": 24555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1613160556195048e-06,
+      "loss": 0.2825,
+      "step": 24556
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1607919120127687e-06,
+      "loss": 0.3498,
+      "step": 24557
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1602678794269817e-06,
+      "loss": 0.3295,
+      "step": 24558
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1597439578687263e-06,
+      "loss": 0.3189,
+      "step": 24559
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1592201473445798e-06,
+      "loss": 0.3048,
+      "step": 24560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1586964478611262e-06,
+      "loss": 0.2956,
+      "step": 24561
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1581728594249375e-06,
+      "loss": 0.3247,
+      "step": 24562
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1576493820425928e-06,
+      "loss": 0.289,
+      "step": 24563
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1571260157206687e-06,
+      "loss": 0.3221,
+      "step": 24564
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.156602760465736e-06,
+      "loss": 0.2838,
+      "step": 24565
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.156079616284368e-06,
+      "loss": 0.3021,
+      "step": 24566
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.155556583183136e-06,
+      "loss": 0.2812,
+      "step": 24567
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1550336611686074e-06,
+      "loss": 0.2989,
+      "step": 24568
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1545108502473545e-06,
+      "loss": 0.3018,
+      "step": 24569
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1539881504259376e-06,
+      "loss": 0.2769,
+      "step": 24570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.153465561710927e-06,
+      "loss": 0.303,
+      "step": 24571
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.152943084108883e-06,
+      "loss": 0.325,
+      "step": 24572
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1524207176263701e-06,
+      "loss": 0.2982,
+      "step": 24573
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151898462269947e-06,
+      "loss": 0.3132,
+      "step": 24574
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1513763180461758e-06,
+      "loss": 0.3361,
+      "step": 24575
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1508542849616122e-06,
+      "loss": 0.3061,
+      "step": 24576
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.150332363022817e-06,
+      "loss": 0.3342,
+      "step": 24577
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1498105522363412e-06,
+      "loss": 0.3103,
+      "step": 24578
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149288852608743e-06,
+      "loss": 0.3147,
+      "step": 24579
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1487672641465697e-06,
+      "loss": 0.3176,
+      "step": 24580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1482457868563767e-06,
+      "loss": 0.2929,
+      "step": 24581
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1477244207447113e-06,
+      "loss": 0.2937,
+      "step": 24582
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.147203165818125e-06,
+      "loss": 0.2954,
+      "step": 24583
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.146682022083161e-06,
+      "loss": 0.2952,
+      "step": 24584
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1461609895463654e-06,
+      "loss": 0.3166,
+      "step": 24585
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1456400682142866e-06,
+      "loss": 0.3126,
+      "step": 24586
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1451192580934623e-06,
+      "loss": 0.3244,
+      "step": 24587
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1445985591904363e-06,
+      "loss": 0.2834,
+      "step": 24588
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.14407797151175e-06,
+      "loss": 0.2998,
+      "step": 24589
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1435574950639406e-06,
+      "loss": 0.291,
+      "step": 24590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1430371298535426e-06,
+      "loss": 0.2891,
+      "step": 24591
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1425168758870963e-06,
+      "loss": 0.3398,
+      "step": 24592
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1419967331711324e-06,
+      "loss": 0.284,
+      "step": 24593
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1414767017121864e-06,
+      "loss": 0.3446,
+      "step": 24594
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1409567815167898e-06,
+      "loss": 0.3104,
+      "step": 24595
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.140436972591471e-06,
+      "loss": 0.3207,
+      "step": 24596
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.13991727494276e-06,
+      "loss": 0.3013,
+      "step": 24597
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1393976885771862e-06,
+      "loss": 0.2946,
+      "step": 24598
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.138878213501271e-06,
+      "loss": 0.2666,
+      "step": 24599
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1383588497215447e-06,
+      "loss": 0.3095,
+      "step": 24600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1378395972445243e-06,
+      "loss": 0.2935,
+      "step": 24601
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1373204560767382e-06,
+      "loss": 0.281,
+      "step": 24602
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1368014262247006e-06,
+      "loss": 0.2796,
+      "step": 24603
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.136282507694937e-06,
+      "loss": 0.3362,
+      "step": 24604
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.135763700493958e-06,
+      "loss": 0.3207,
+      "step": 24605
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1352450046282837e-06,
+      "loss": 0.3184,
+      "step": 24606
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1347264201044305e-06,
+      "loss": 0.3008,
+      "step": 24607
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1342079469289081e-06,
+      "loss": 0.2983,
+      "step": 24608
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1336895851082308e-06,
+      "loss": 0.2888,
+      "step": 24609
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1331713346489105e-06,
+      "loss": 0.3153,
+      "step": 24610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1326531955574526e-06,
+      "loss": 0.3276,
+      "step": 24611
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1321351678403703e-06,
+      "loss": 0.2834,
+      "step": 24612
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1316172515041634e-06,
+      "loss": 0.2683,
+      "step": 24613
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1310994465553438e-06,
+      "loss": 0.2926,
+      "step": 24614
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1305817530004093e-06,
+      "loss": 0.3273,
+      "step": 24615
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.130064170845865e-06,
+      "loss": 0.3015,
+      "step": 24616
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.129546700098213e-06,
+      "loss": 0.3221,
+      "step": 24617
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1290293407639496e-06,
+      "loss": 0.2775,
+      "step": 24618
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1285120928495742e-06,
+      "loss": 0.3049,
+      "step": 24619
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1279949563615867e-06,
+      "loss": 0.3244,
+      "step": 24620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1274779313064765e-06,
+      "loss": 0.309,
+      "step": 24621
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.126961017690742e-06,
+      "loss": 0.3174,
+      "step": 24622
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1264442155208754e-06,
+      "loss": 0.3066,
+      "step": 24623
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1259275248033629e-06,
+      "loss": 0.3258,
+      "step": 24624
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1254109455446994e-06,
+      "loss": 0.3063,
+      "step": 24625
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1248944777513726e-06,
+      "loss": 0.3205,
+      "step": 24626
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1243781214298655e-06,
+      "loss": 0.2983,
+      "step": 24627
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.123861876586667e-06,
+      "loss": 0.327,
+      "step": 24628
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1233457432282624e-06,
+      "loss": 0.3232,
+      "step": 24629
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1228297213611306e-06,
+      "loss": 0.2988,
+      "step": 24630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1223138109917542e-06,
+      "loss": 0.31,
+      "step": 24631
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.121798012126616e-06,
+      "loss": 0.2827,
+      "step": 24632
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1212823247721927e-06,
+      "loss": 0.3378,
+      "step": 24633
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.120766748934957e-06,
+      "loss": 0.2966,
+      "step": 24634
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.120251284621391e-06,
+      "loss": 0.3314,
+      "step": 24635
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1197359318379642e-06,
+      "loss": 0.3197,
+      "step": 24636
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1192206905911518e-06,
+      "loss": 0.3308,
+      "step": 24637
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1187055608874265e-06,
+      "loss": 0.2971,
+      "step": 24638
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1181905427332552e-06,
+      "loss": 0.2992,
+      "step": 24639
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.117675636135107e-06,
+      "loss": 0.3226,
+      "step": 24640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.117160841099454e-06,
+      "loss": 0.3419,
+      "step": 24641
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1166461576327546e-06,
+      "loss": 0.3443,
+      "step": 24642
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.11613158574148e-06,
+      "loss": 0.2928,
+      "step": 24643
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1156171254320903e-06,
+      "loss": 0.3141,
+      "step": 24644
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1151027767110446e-06,
+      "loss": 0.3178,
+      "step": 24645
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1145885395848054e-06,
+      "loss": 0.2995,
+      "step": 24646
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1140744140598325e-06,
+      "loss": 0.2717,
+      "step": 24647
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1135604001425848e-06,
+      "loss": 0.3415,
+      "step": 24648
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1130464978395138e-06,
+      "loss": 0.3017,
+      "step": 24649
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1125327071570758e-06,
+      "loss": 0.2971,
+      "step": 24650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1120190281017273e-06,
+      "loss": 0.3204,
+      "step": 24651
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1115054606799159e-06,
+      "loss": 0.2875,
+      "step": 24652
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.110992004898096e-06,
+      "loss": 0.3296,
+      "step": 24653
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1104786607627139e-06,
+      "loss": 0.3377,
+      "step": 24654
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1099654282802154e-06,
+      "loss": 0.2974,
+      "step": 24655
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1094523074570484e-06,
+      "loss": 0.3122,
+      "step": 24656
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1089392982996615e-06,
+      "loss": 0.2714,
+      "step": 24657
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.108426400814493e-06,
+      "loss": 0.3021,
+      "step": 24658
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1079136150079862e-06,
+      "loss": 0.3093,
+      "step": 24659
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.107400940886585e-06,
+      "loss": 0.2847,
+      "step": 24660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1068883784567241e-06,
+      "loss": 0.3008,
+      "step": 24661
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.106375927724842e-06,
+      "loss": 0.2964,
+      "step": 24662
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1058635886973801e-06,
+      "loss": 0.3083,
+      "step": 24663
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1053513613807687e-06,
+      "loss": 0.2926,
+      "step": 24664
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1048392457814406e-06,
+      "loss": 0.3248,
+      "step": 24665
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1043272419058314e-06,
+      "loss": 0.3102,
+      "step": 24666
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1038153497603688e-06,
+      "loss": 0.3069,
+      "step": 24667
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1033035693514827e-06,
+      "loss": 0.3067,
+      "step": 24668
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1027919006856047e-06,
+      "loss": 0.2804,
+      "step": 24669
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1022803437691565e-06,
+      "loss": 0.3252,
+      "step": 24670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.101768898608566e-06,
+      "loss": 0.2929,
+      "step": 24671
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1012575652102586e-06,
+      "loss": 0.3062,
+      "step": 24672
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100746343580652e-06,
+      "loss": 0.3033,
+      "step": 24673
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1002352337261712e-06,
+      "loss": 0.2904,
+      "step": 24674
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0997242356532335e-06,
+      "loss": 0.2928,
+      "step": 24675
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0992133493682599e-06,
+      "loss": 0.3065,
+      "step": 24676
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0987025748776625e-06,
+      "loss": 0.3454,
+      "step": 24677
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0981919121878615e-06,
+      "loss": 0.3204,
+      "step": 24678
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0976813613052662e-06,
+      "loss": 0.3098,
+      "step": 24679
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0971709222362913e-06,
+      "loss": 0.3141,
+      "step": 24680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.096660594987351e-06,
+      "loss": 0.323,
+      "step": 24681
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.096150379564851e-06,
+      "loss": 0.3244,
+      "step": 24682
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0956402759751994e-06,
+      "loss": 0.2954,
+      "step": 24683
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.095130284224808e-06,
+      "loss": 0.2687,
+      "step": 24684
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094620404320076e-06,
+      "loss": 0.2739,
+      "step": 24685
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0941106362674126e-06,
+      "loss": 0.3086,
+      "step": 24686
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0936009800732162e-06,
+      "loss": 0.3135,
+      "step": 24687
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0930914357438926e-06,
+      "loss": 0.3049,
+      "step": 24688
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.092582003285837e-06,
+      "loss": 0.3224,
+      "step": 24689
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0920726827054506e-06,
+      "loss": 0.323,
+      "step": 24690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.091563474009133e-06,
+      "loss": 0.2832,
+      "step": 24691
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0910543772032744e-06,
+      "loss": 0.3243,
+      "step": 24692
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.090545392294272e-06,
+      "loss": 0.3176,
+      "step": 24693
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.09003651928852e-06,
+      "loss": 0.2967,
+      "step": 24694
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.089527758192407e-06,
+      "loss": 0.3265,
+      "step": 24695
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0890191090123259e-06,
+      "loss": 0.3322,
+      "step": 24696
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.088510571754664e-06,
+      "loss": 0.3199,
+      "step": 24697
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0880021464258062e-06,
+      "loss": 0.3288,
+      "step": 24698
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0874938330321406e-06,
+      "loss": 0.316,
+      "step": 24699
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.086985631580053e-06,
+      "loss": 0.2808,
+      "step": 24700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0864775420759244e-06,
+      "loss": 0.3223,
+      "step": 24701
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.085969564526137e-06,
+      "loss": 0.3124,
+      "step": 24702
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0854616989370725e-06,
+      "loss": 0.3226,
+      "step": 24703
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0849539453151059e-06,
+      "loss": 0.2884,
+      "step": 24704
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0844463036666176e-06,
+      "loss": 0.3201,
+      "step": 24705
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0839387739979857e-06,
+      "loss": 0.3152,
+      "step": 24706
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0834313563155807e-06,
+      "loss": 0.2979,
+      "step": 24707
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0829240506257765e-06,
+      "loss": 0.3037,
+      "step": 24708
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0824168569349469e-06,
+      "loss": 0.3114,
+      "step": 24709
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.081909775249459e-06,
+      "loss": 0.3456,
+      "step": 24710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0814028055756843e-06,
+      "loss": 0.3019,
+      "step": 24711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.080895947919992e-06,
+      "loss": 0.3237,
+      "step": 24712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.080389202288743e-06,
+      "loss": 0.2779,
+      "step": 24713
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0798825686883063e-06,
+      "loss": 0.3411,
+      "step": 24714
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0793760471250458e-06,
+      "loss": 0.3104,
+      "step": 24715
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0788696376053198e-06,
+      "loss": 0.2932,
+      "step": 24716
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.078363340135492e-06,
+      "loss": 0.337,
+      "step": 24717
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0778571547219218e-06,
+      "loss": 0.324,
+      "step": 24718
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.077351081370962e-06,
+      "loss": 0.2841,
+      "step": 24719
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.076845120088973e-06,
+      "loss": 0.3025,
+      "step": 24720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0763392708823094e-06,
+      "loss": 0.3137,
+      "step": 24721
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0758335337573255e-06,
+      "loss": 0.3166,
+      "step": 24722
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0753279087203716e-06,
+      "loss": 0.2844,
+      "step": 24723
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.074822395777798e-06,
+      "loss": 0.3282,
+      "step": 24724
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0743169949359566e-06,
+      "loss": 0.2895,
+      "step": 24725
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.073811706201192e-06,
+      "loss": 0.3213,
+      "step": 24726
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.073306529579855e-06,
+      "loss": 0.3171,
+      "step": 24727
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072801465078287e-06,
+      "loss": 0.3013,
+      "step": 24728
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0722965127028306e-06,
+      "loss": 0.2703,
+      "step": 24729
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.071791672459831e-06,
+      "loss": 0.2811,
+      "step": 24730
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0712869443556284e-06,
+      "loss": 0.2927,
+      "step": 24731
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0707823283965612e-06,
+      "loss": 0.3467,
+      "step": 24732
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0702778245889678e-06,
+      "loss": 0.3422,
+      "step": 24733
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0697734329391873e-06,
+      "loss": 0.2959,
+      "step": 24734
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0692691534535494e-06,
+      "loss": 0.3238,
+      "step": 24735
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068764986138392e-06,
+      "loss": 0.287,
+      "step": 24736
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0682609310000492e-06,
+      "loss": 0.2891,
+      "step": 24737
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0677569880448479e-06,
+      "loss": 0.3191,
+      "step": 24738
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0672531572791178e-06,
+      "loss": 0.2921,
+      "step": 24739
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0667494387091903e-06,
+      "loss": 0.3569,
+      "step": 24740
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0662458323413882e-06,
+      "loss": 0.3226,
+      "step": 24741
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0657423381820397e-06,
+      "loss": 0.3319,
+      "step": 24742
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0652389562374687e-06,
+      "loss": 0.3386,
+      "step": 24743
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0647356865139946e-06,
+      "loss": 0.2966,
+      "step": 24744
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.064232529017941e-06,
+      "loss": 0.2676,
+      "step": 24745
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.06372948375563e-06,
+      "loss": 0.3002,
+      "step": 24746
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.063226550733375e-06,
+      "loss": 0.3144,
+      "step": 24747
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0627237299574966e-06,
+      "loss": 0.3145,
+      "step": 24748
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0622210214343086e-06,
+      "loss": 0.3562,
+      "step": 24749
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.061718425170123e-06,
+      "loss": 0.3416,
+      "step": 24750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0612159411712553e-06,
+      "loss": 0.2759,
+      "step": 24751
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0607135694440175e-06,
+      "loss": 0.2991,
+      "step": 24752
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0602113099947153e-06,
+      "loss": 0.2839,
+      "step": 24753
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0597091628296607e-06,
+      "loss": 0.3267,
+      "step": 24754
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0592071279551619e-06,
+      "loss": 0.3327,
+      "step": 24755
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0587052053775193e-06,
+      "loss": 0.2948,
+      "step": 24756
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0582033951030414e-06,
+      "loss": 0.3066,
+      "step": 24757
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0577016971380305e-06,
+      "loss": 0.3089,
+      "step": 24758
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0572001114887875e-06,
+      "loss": 0.2844,
+      "step": 24759
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0566986381616096e-06,
+      "loss": 0.3117,
+      "step": 24760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0561972771627982e-06,
+      "loss": 0.3222,
+      "step": 24761
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0556960284986518e-06,
+      "loss": 0.3049,
+      "step": 24762
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0551948921754628e-06,
+      "loss": 0.3137,
+      "step": 24763
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0546938681995256e-06,
+      "loss": 0.3029,
+      "step": 24764
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.054192956577137e-06,
+      "loss": 0.3271,
+      "step": 24765
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0536921573145842e-06,
+      "loss": 0.3204,
+      "step": 24766
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0531914704181589e-06,
+      "loss": 0.3108,
+      "step": 24767
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0526908958941517e-06,
+      "loss": 0.32,
+      "step": 24768
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0521904337488476e-06,
+      "loss": 0.3202,
+      "step": 24769
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0516900839885314e-06,
+      "loss": 0.284,
+      "step": 24770
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0511898466194903e-06,
+      "loss": 0.3007,
+      "step": 24771
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050689721648005e-06,
+      "loss": 0.2803,
+      "step": 24772
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050189709080357e-06,
+      "loss": 0.3062,
+      "step": 24773
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.049689808922829e-06,
+      "loss": 0.2915,
+      "step": 24774
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0491900211816974e-06,
+      "loss": 0.3101,
+      "step": 24775
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0486903458632403e-06,
+      "loss": 0.291,
+      "step": 24776
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.048190782973736e-06,
+      "loss": 0.3066,
+      "step": 24777
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0476913325194538e-06,
+      "loss": 0.3579,
+      "step": 24778
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0471919945066721e-06,
+      "loss": 0.2783,
+      "step": 24779
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0466927689416595e-06,
+      "loss": 0.32,
+      "step": 24780
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461936558306873e-06,
+      "loss": 0.3389,
+      "step": 24781
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.045694655180024e-06,
+      "loss": 0.3121,
+      "step": 24782
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0451957669959389e-06,
+      "loss": 0.3155,
+      "step": 24783
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0446969912846938e-06,
+      "loss": 0.3179,
+      "step": 24784
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0441983280525569e-06,
+      "loss": 0.3052,
+      "step": 24785
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0436997773057912e-06,
+      "loss": 0.306,
+      "step": 24786
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0432013390506567e-06,
+      "loss": 0.3154,
+      "step": 24787
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0427030132934158e-06,
+      "loss": 0.3273,
+      "step": 24788
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0422048000403285e-06,
+      "loss": 0.2995,
+      "step": 24789
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0417066992976478e-06,
+      "loss": 0.3032,
+      "step": 24790
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.041208711071635e-06,
+      "loss": 0.3159,
+      "step": 24791
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0407108353685424e-06,
+      "loss": 0.3183,
+      "step": 24792
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0402130721946213e-06,
+      "loss": 0.3176,
+      "step": 24793
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0397154215561268e-06,
+      "loss": 0.3411,
+      "step": 24794
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0392178834593082e-06,
+      "loss": 0.3099,
+      "step": 24795
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0387204579104159e-06,
+      "loss": 0.2855,
+      "step": 24796
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0382231449156955e-06,
+      "loss": 0.3145,
+      "step": 24797
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0377259444813937e-06,
+      "loss": 0.2892,
+      "step": 24798
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0372288566137578e-06,
+      "loss": 0.3126,
+      "step": 24799
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0367318813190285e-06,
+      "loss": 0.2857,
+      "step": 24800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0362350186034497e-06,
+      "loss": 0.3074,
+      "step": 24801
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0357382684732619e-06,
+      "loss": 0.3234,
+      "step": 24802
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0352416309347003e-06,
+      "loss": 0.3287,
+      "step": 24803
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0347451059940072e-06,
+      "loss": 0.3055,
+      "step": 24804
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0342486936574192e-06,
+      "loss": 0.3099,
+      "step": 24805
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0337523939311678e-06,
+      "loss": 0.2736,
+      "step": 24806
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.033256206821488e-06,
+      "loss": 0.2982,
+      "step": 24807
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0327601323346147e-06,
+      "loss": 0.3103,
+      "step": 24808
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0322641704767744e-06,
+      "loss": 0.2872,
+      "step": 24809
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0317683212541984e-06,
+      "loss": 0.2924,
+      "step": 24810
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0312725846731174e-06,
+      "loss": 0.3013,
+      "step": 24811
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0307769607397543e-06,
+      "loss": 0.2841,
+      "step": 24812
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.030281449460333e-06,
+      "loss": 0.3104,
+      "step": 24813
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0297860508410807e-06,
+      "loss": 0.3099,
+      "step": 24814
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0292907648882167e-06,
+      "loss": 0.3231,
+      "step": 24815
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0287955916079617e-06,
+      "loss": 0.3119,
+      "step": 24816
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0283005310065396e-06,
+      "loss": 0.3356,
+      "step": 24817
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0278055830901634e-06,
+      "loss": 0.2888,
+      "step": 24818
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0273107478650513e-06,
+      "loss": 0.3276,
+      "step": 24819
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0268160253374205e-06,
+      "loss": 0.3309,
+      "step": 24820
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0263214155134814e-06,
+      "loss": 0.3036,
+      "step": 24821
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0258269183994495e-06,
+      "loss": 0.2985,
+      "step": 24822
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0253325340015353e-06,
+      "loss": 0.2979,
+      "step": 24823
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0248382623259434e-06,
+      "loss": 0.3277,
+      "step": 24824
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.024344103378887e-06,
+      "loss": 0.3068,
+      "step": 24825
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0238500571665723e-06,
+      "loss": 0.2918,
+      "step": 24826
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0233561236952016e-06,
+      "loss": 0.3418,
+      "step": 24827
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0228623029709816e-06,
+      "loss": 0.3342,
+      "step": 24828
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.022368595000115e-06,
+      "loss": 0.3125,
+      "step": 24829
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0218749997888e-06,
+      "loss": 0.3123,
+      "step": 24830
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0213815173432383e-06,
+      "loss": 0.3009,
+      "step": 24831
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0208881476696297e-06,
+      "loss": 0.3424,
+      "step": 24832
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0203948907741679e-06,
+      "loss": 0.3232,
+      "step": 24833
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0199017466630478e-06,
+      "loss": 0.301,
+      "step": 24834
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0194087153424648e-06,
+      "loss": 0.3294,
+      "step": 24835
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0189157968186125e-06,
+      "loss": 0.3068,
+      "step": 24836
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0184229910976796e-06,
+      "loss": 0.3037,
+      "step": 24837
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0179302981858562e-06,
+      "loss": 0.2832,
+      "step": 24838
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0174377180893336e-06,
+      "loss": 0.325,
+      "step": 24839
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0169452508142951e-06,
+      "loss": 0.3324,
+      "step": 24840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0164528963669261e-06,
+      "loss": 0.3362,
+      "step": 24841
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0159606547534151e-06,
+      "loss": 0.3205,
+      "step": 24842
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0154685259799412e-06,
+      "loss": 0.2679,
+      "step": 24843
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.014976510052683e-06,
+      "loss": 0.3016,
+      "step": 24844
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0144846069778259e-06,
+      "loss": 0.292,
+      "step": 24845
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0139928167615431e-06,
+      "loss": 0.2828,
+      "step": 24846
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0135011394100136e-06,
+      "loss": 0.3121,
+      "step": 24847
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0130095749294155e-06,
+      "loss": 0.2779,
+      "step": 24848
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0125181233259174e-06,
+      "loss": 0.3252,
+      "step": 24849
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0120267846056963e-06,
+      "loss": 0.3138,
+      "step": 24850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0115355587749231e-06,
+      "loss": 0.3332,
+      "step": 24851
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0110444458397662e-06,
+      "loss": 0.2981,
+      "step": 24852
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0105534458063948e-06,
+      "loss": 0.309,
+      "step": 24853
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0100625586809764e-06,
+      "loss": 0.3017,
+      "step": 24854
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.009571784469674e-06,
+      "loss": 0.264,
+      "step": 24855
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0090811231786535e-06,
+      "loss": 0.292,
+      "step": 24856
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.008590574814079e-06,
+      "loss": 0.3038,
+      "step": 24857
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0081001393821087e-06,
+      "loss": 0.2693,
+      "step": 24858
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0076098168889036e-06,
+      "loss": 0.3114,
+      "step": 24859
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0071196073406264e-06,
+      "loss": 0.3022,
+      "step": 24860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0066295107434276e-06,
+      "loss": 0.316,
+      "step": 24861
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0061395271034658e-06,
+      "loss": 0.3091,
+      "step": 24862
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0056496564268981e-06,
+      "loss": 0.3222,
+      "step": 24863
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.005159898719873e-06,
+      "loss": 0.3244,
+      "step": 24864
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0046702539885422e-06,
+      "loss": 0.2821,
+      "step": 24865
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0041807222390564e-06,
+      "loss": 0.2896,
+      "step": 24866
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0036913034775675e-06,
+      "loss": 0.3181,
+      "step": 24867
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.003201997710217e-06,
+      "loss": 0.3045,
+      "step": 24868
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0027128049431534e-06,
+      "loss": 0.2973,
+      "step": 24869
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0022237251825228e-06,
+      "loss": 0.3269,
+      "step": 24870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.001734758434464e-06,
+      "loss": 0.3006,
+      "step": 24871
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0012459047051216e-06,
+      "loss": 0.3213,
+      "step": 24872
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0007571640006353e-06,
+      "loss": 0.2943,
+      "step": 24873
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0002685363271436e-06,
+      "loss": 0.3348,
+      "step": 24874
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.997800216907806e-07,
+      "loss": 0.3136,
+      "step": 24875
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992916200976877e-07,
+      "loss": 0.2851,
+      "step": 24876
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988033315539936e-07,
+      "loss": 0.3296,
+      "step": 24877
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983151560658333e-07,
+      "loss": 0.316,
+      "step": 24878
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.978270936393408e-07,
+      "loss": 0.3434,
+      "step": 24879
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.973391442806424e-07,
+      "loss": 0.2966,
+      "step": 24880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.968513079958698e-07,
+      "loss": 0.3073,
+      "step": 24881
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96363584791149e-07,
+      "loss": 0.3125,
+      "step": 24882
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958759746726055e-07,
+      "loss": 0.3087,
+      "step": 24883
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.953884776463652e-07,
+      "loss": 0.3197,
+      "step": 24884
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.949010937185476e-07,
+      "loss": 0.3064,
+      "step": 24885
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.944138228952792e-07,
+      "loss": 0.279,
+      "step": 24886
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.939266651826751e-07,
+      "loss": 0.3381,
+      "step": 24887
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93439620586859e-07,
+      "loss": 0.2693,
+      "step": 24888
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.92952689113943e-07,
+      "loss": 0.3354,
+      "step": 24889
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.924658707700462e-07,
+      "loss": 0.3131,
+      "step": 24890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.919791655612842e-07,
+      "loss": 0.2928,
+      "step": 24891
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.914925734937675e-07,
+      "loss": 0.3436,
+      "step": 24892
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.910060945736077e-07,
+      "loss": 0.2839,
+      "step": 24893
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.90519728806918e-07,
+      "loss": 0.3452,
+      "step": 24894
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.900334761998031e-07,
+      "loss": 0.2999,
+      "step": 24895
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.895473367583742e-07,
+      "loss": 0.3014,
+      "step": 24896
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.890613104887349e-07,
+      "loss": 0.3365,
+      "step": 24897
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.885753973969893e-07,
+      "loss": 0.3195,
+      "step": 24898
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.880895974892413e-07,
+      "loss": 0.3246,
+      "step": 24899
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.87603910771594e-07,
+      "loss": 0.3284,
+      "step": 24900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.871183372501447e-07,
+      "loss": 0.2947,
+      "step": 24901
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.86632876930994e-07,
+      "loss": 0.3201,
+      "step": 24902
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.861475298202382e-07,
+      "loss": 0.2976,
+      "step": 24903
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.85662295923977e-07,
+      "loss": 0.318,
+      "step": 24904
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.851771752483007e-07,
+      "loss": 0.2837,
+      "step": 24905
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.84692167799305e-07,
+      "loss": 0.2961,
+      "step": 24906
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.84207273583081e-07,
+      "loss": 0.2858,
+      "step": 24907
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.837224926057176e-07,
+      "loss": 0.3029,
+      "step": 24908
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.832378248733032e-07,
+      "loss": 0.3083,
+      "step": 24909
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.827532703919296e-07,
+      "loss": 0.3069,
+      "step": 24910
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.822688291676774e-07,
+      "loss": 0.2838,
+      "step": 24911
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.81784501206634e-07,
+      "loss": 0.3291,
+      "step": 24912
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.813002865148847e-07,
+      "loss": 0.3062,
+      "step": 24913
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.808161850985065e-07,
+      "loss": 0.3265,
+      "step": 24914
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.803321969635816e-07,
+      "loss": 0.3059,
+      "step": 24915
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.798483221161925e-07,
+      "loss": 0.2918,
+      "step": 24916
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.793645605624124e-07,
+      "loss": 0.3059,
+      "step": 24917
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.788809123083175e-07,
+      "loss": 0.3408,
+      "step": 24918
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.783973773599853e-07,
+      "loss": 0.3155,
+      "step": 24919
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.779139557234863e-07,
+      "loss": 0.2971,
+      "step": 24920
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.774306474048922e-07,
+      "loss": 0.3022,
+      "step": 24921
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.769474524102772e-07,
+      "loss": 0.2391,
+      "step": 24922
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.764643707457045e-07,
+      "loss": 0.2978,
+      "step": 24923
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.759814024172454e-07,
+      "loss": 0.302,
+      "step": 24924
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.754985474309675e-07,
+      "loss": 0.3062,
+      "step": 24925
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.750158057929304e-07,
+      "loss": 0.2977,
+      "step": 24926
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.745331775092037e-07,
+      "loss": 0.3074,
+      "step": 24927
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.740506625858447e-07,
+      "loss": 0.3103,
+      "step": 24928
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.735682610289133e-07,
+      "loss": 0.2816,
+      "step": 24929
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.730859728444696e-07,
+      "loss": 0.3132,
+      "step": 24930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.72603798038574e-07,
+      "loss": 0.2972,
+      "step": 24931
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.721217366172774e-07,
+      "loss": 0.2768,
+      "step": 24932
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.716397885866392e-07,
+      "loss": 0.28,
+      "step": 24933
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.711579539527117e-07,
+      "loss": 0.3238,
+      "step": 24934
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.706762327215436e-07,
+      "loss": 0.3277,
+      "step": 24935
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.701946248991888e-07,
+      "loss": 0.291,
+      "step": 24936
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.697131304916963e-07,
+      "loss": 0.2691,
+      "step": 24937
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.69231749505113e-07,
+      "loss": 0.2707,
+      "step": 24938
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.68750481945483e-07,
+      "loss": 0.3175,
+      "step": 24939
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.682693278188527e-07,
+      "loss": 0.3105,
+      "step": 24940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.67788287131267e-07,
+      "loss": 0.3344,
+      "step": 24941
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.67307359888765e-07,
+      "loss": 0.3404,
+      "step": 24942
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66826546097388e-07,
+      "loss": 0.2892,
+      "step": 24943
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.663458457631768e-07,
+      "loss": 0.3242,
+      "step": 24944
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.658652588921668e-07,
+      "loss": 0.3057,
+      "step": 24945
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.653847854903943e-07,
+      "loss": 0.286,
+      "step": 24946
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.649044255638973e-07,
+      "loss": 0.3169,
+      "step": 24947
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.64424179118707e-07,
+      "loss": 0.2829,
+      "step": 24948
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63944046160853e-07,
+      "loss": 0.2945,
+      "step": 24949
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63464026696369e-07,
+      "loss": 0.3104,
+      "step": 24950
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.629841207312818e-07,
+      "loss": 0.275,
+      "step": 24951
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.625043282716207e-07,
+      "loss": 0.295,
+      "step": 24952
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.62024649323412e-07,
+      "loss": 0.3105,
+      "step": 24953
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.615450838926787e-07,
+      "loss": 0.322,
+      "step": 24954
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61065631985445e-07,
+      "loss": 0.2982,
+      "step": 24955
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.605862936077359e-07,
+      "loss": 0.2814,
+      "step": 24956
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.601070687655667e-07,
+      "loss": 0.3312,
+      "step": 24957
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.596279574649603e-07,
+      "loss": 0.3425,
+      "step": 24958
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59148959711932e-07,
+      "loss": 0.3252,
+      "step": 24959
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.58670075512501e-07,
+      "loss": 0.3154,
+      "step": 24960
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.581913048726788e-07,
+      "loss": 0.3085,
+      "step": 24961
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.57712647798481e-07,
+      "loss": 0.329,
+      "step": 24962
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.572341042959177e-07,
+      "loss": 0.2983,
+      "step": 24963
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.567556743709994e-07,
+      "loss": 0.2989,
+      "step": 24964
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.56277358029739e-07,
+      "loss": 0.3043,
+      "step": 24965
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.557991552781398e-07,
+      "loss": 0.3055,
+      "step": 24966
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.5532106612221e-07,
+      "loss": 0.3195,
+      "step": 24967
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54843090567955e-07,
+      "loss": 0.3115,
+      "step": 24968
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.543652286213768e-07,
+      "loss": 0.3525,
+      "step": 24969
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.53887480288479e-07,
+      "loss": 0.3198,
+      "step": 24970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.534098455752605e-07,
+      "loss": 0.3016,
+      "step": 24971
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.529323244877197e-07,
+      "loss": 0.3094,
+      "step": 24972
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.524549170318554e-07,
+      "loss": 0.3098,
+      "step": 24973
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.519776232136668e-07,
+      "loss": 0.3356,
+      "step": 24974
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.515004430391439e-07,
+      "loss": 0.2789,
+      "step": 24975
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.510233765142817e-07,
+      "loss": 0.3033,
+      "step": 24976
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.505464236450723e-07,
+      "loss": 0.3226,
+      "step": 24977
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.500695844375097e-07,
+      "loss": 0.2617,
+      "step": 24978
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.495928588975767e-07,
+      "loss": 0.3057,
+      "step": 24979
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.491162470312676e-07,
+      "loss": 0.3008,
+      "step": 24980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.486397488445643e-07,
+      "loss": 0.3049,
+      "step": 24981
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481633643434506e-07,
+      "loss": 0.2995,
+      "step": 24982
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.476870935339133e-07,
+      "loss": 0.2872,
+      "step": 24983
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.47210936421934e-07,
+      "loss": 0.3141,
+      "step": 24984
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.467348930134901e-07,
+      "loss": 0.3331,
+      "step": 24985
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.462589633145635e-07,
+      "loss": 0.2982,
+      "step": 24986
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.457831473311341e-07,
+      "loss": 0.2725,
+      "step": 24987
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.453074450691724e-07,
+      "loss": 0.3005,
+      "step": 24988
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.44831856534657e-07,
+      "loss": 0.3123,
+      "step": 24989
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.443563817335622e-07,
+      "loss": 0.3174,
+      "step": 24990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.438810206718585e-07,
+      "loss": 0.3207,
+      "step": 24991
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.434057733555147e-07,
+      "loss": 0.3225,
+      "step": 24992
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.429306397905025e-07,
+      "loss": 0.2963,
+      "step": 24993
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.424556199827883e-07,
+      "loss": 0.2968,
+      "step": 24994
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.419807139383374e-07,
+      "loss": 0.3007,
+      "step": 24995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.415059216631184e-07,
+      "loss": 0.3049,
+      "step": 24996
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.410312431630908e-07,
+      "loss": 0.3393,
+      "step": 24997
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.405566784442188e-07,
+      "loss": 0.2933,
+      "step": 24998
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.400822275124632e-07,
+      "loss": 0.3541,
+      "step": 24999
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.396078903737804e-07,
+      "loss": 0.2899,
+      "step": 25000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.391336670341322e-07,
+      "loss": 0.2919,
+      "step": 25001
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.386595574994717e-07,
+      "loss": 0.2915,
+      "step": 25002
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.381855617757541e-07,
+      "loss": 0.3074,
+      "step": 25003
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.377116798689323e-07,
+      "loss": 0.3209,
+      "step": 25004
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.372379117849617e-07,
+      "loss": 0.2974,
+      "step": 25005
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.367642575297886e-07,
+      "loss": 0.2877,
+      "step": 25006
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.362907171093638e-07,
+      "loss": 0.3258,
+      "step": 25007
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.35817290529637e-07,
+      "loss": 0.3622,
+      "step": 25008
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.353439777965512e-07,
+      "loss": 0.3139,
+      "step": 25009
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.348707789160527e-07,
+      "loss": 0.31,
+      "step": 25010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.343976938940869e-07,
+      "loss": 0.2867,
+      "step": 25011
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.339247227365933e-07,
+      "loss": 0.303,
+      "step": 25012
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334518654495117e-07,
+      "loss": 0.3238,
+      "step": 25013
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.329791220387818e-07,
+      "loss": 0.2705,
+      "step": 25014
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.325064925103445e-07,
+      "loss": 0.2918,
+      "step": 25015
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.320339768701315e-07,
+      "loss": 0.3057,
+      "step": 25016
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.315615751240792e-07,
+      "loss": 0.3123,
+      "step": 25017
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.31089287278123e-07,
+      "loss": 0.2791,
+      "step": 25018
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.306171133381914e-07,
+      "loss": 0.309,
+      "step": 25019
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.301450533102174e-07,
+      "loss": 0.3154,
+      "step": 25020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.296731072001319e-07,
+      "loss": 0.2479,
+      "step": 25021
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.29201275013859e-07,
+      "loss": 0.3296,
+      "step": 25022
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.287295567573251e-07,
+      "loss": 0.3243,
+      "step": 25023
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.282579524364577e-07,
+      "loss": 0.3032,
+      "step": 25024
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.277864620571753e-07,
+      "loss": 0.284,
+      "step": 25025
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.273150856254043e-07,
+      "loss": 0.3222,
+      "step": 25026
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.268438231470656e-07,
+      "loss": 0.3343,
+      "step": 25027
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.263726746280755e-07,
+      "loss": 0.2654,
+      "step": 25028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.259016400743515e-07,
+      "loss": 0.3202,
+      "step": 25029
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254307194918144e-07,
+      "loss": 0.3474,
+      "step": 25030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.249599128863729e-07,
+      "loss": 0.2857,
+      "step": 25031
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.244892202639444e-07,
+      "loss": 0.278,
+      "step": 25032
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.240186416304408e-07,
+      "loss": 0.3162,
+      "step": 25033
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.235481769917687e-07,
+      "loss": 0.3048,
+      "step": 25034
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.230778263538398e-07,
+      "loss": 0.3053,
+      "step": 25035
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.22607589722565e-07,
+      "loss": 0.314,
+      "step": 25036
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.221374671038441e-07,
+      "loss": 0.3127,
+      "step": 25037
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.216674585035857e-07,
+      "loss": 0.318,
+      "step": 25038
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.211975639276938e-07,
+      "loss": 0.3189,
+      "step": 25039
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.207277833820671e-07,
+      "loss": 0.3211,
+      "step": 25040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.202581168726077e-07,
+      "loss": 0.3056,
+      "step": 25041
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.197885644052174e-07,
+      "loss": 0.3154,
+      "step": 25042
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.193191259857903e-07,
+      "loss": 0.2696,
+      "step": 25043
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18849801620223e-07,
+      "loss": 0.3105,
+      "step": 25044
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.183805913144117e-07,
+      "loss": 0.2933,
+      "step": 25045
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.179114950742473e-07,
+      "loss": 0.2926,
+      "step": 25046
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.17442512905623e-07,
+      "loss": 0.3331,
+      "step": 25047
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.169736448144318e-07,
+      "loss": 0.2853,
+      "step": 25048
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.165048908065577e-07,
+      "loss": 0.3052,
+      "step": 25049
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.160362508878917e-07,
+      "loss": 0.2934,
+      "step": 25050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.155677250643191e-07,
+      "loss": 0.288,
+      "step": 25051
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.150993133417263e-07,
+      "loss": 0.34,
+      "step": 25052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.146310157259941e-07,
+      "loss": 0.3384,
+      "step": 25053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.141628322230045e-07,
+      "loss": 0.3062,
+      "step": 25054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.136947628386395e-07,
+      "loss": 0.3131,
+      "step": 25055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.132268075787765e-07,
+      "loss": 0.3221,
+      "step": 25056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.12758966449292e-07,
+      "loss": 0.3071,
+      "step": 25057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.122912394560668e-07,
+      "loss": 0.2994,
+      "step": 25058
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.118236266049707e-07,
+      "loss": 0.2827,
+      "step": 25059
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.11356127901879e-07,
+      "loss": 0.2989,
+      "step": 25060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.108887433526648e-07,
+      "loss": 0.3126,
+      "step": 25061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.104214729631955e-07,
+      "loss": 0.2903,
+      "step": 25062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.099543167393432e-07,
+      "loss": 0.2886,
+      "step": 25063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.09487274686972e-07,
+      "loss": 0.3097,
+      "step": 25064
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.090203468119518e-07,
+      "loss": 0.3186,
+      "step": 25065
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.085535331201433e-07,
+      "loss": 0.3229,
+      "step": 25066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.080868336174131e-07,
+      "loss": 0.2812,
+      "step": 25067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.076202483096197e-07,
+      "loss": 0.3334,
+      "step": 25068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.071537772026251e-07,
+      "loss": 0.3196,
+      "step": 25069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.066874203022913e-07,
+      "loss": 0.3347,
+      "step": 25070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.062211776144691e-07,
+      "loss": 0.2855,
+      "step": 25071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.057550491450195e-07,
+      "loss": 0.3185,
+      "step": 25072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.052890348997978e-07,
+      "loss": 0.3287,
+      "step": 25073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.048231348846526e-07,
+      "loss": 0.2934,
+      "step": 25074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.043573491054403e-07,
+      "loss": 0.3157,
+      "step": 25075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.038916775680095e-07,
+      "loss": 0.3187,
+      "step": 25076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.034261202782057e-07,
+      "loss": 0.3159,
+      "step": 25077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.029606772418809e-07,
+      "loss": 0.3015,
+      "step": 25078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.024953484648802e-07,
+      "loss": 0.3243,
+      "step": 25079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.020301339530457e-07,
+      "loss": 0.2896,
+      "step": 25080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.015650337122229e-07,
+      "loss": 0.3148,
+      "step": 25081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.011000477482534e-07,
+      "loss": 0.3485,
+      "step": 25082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.006351760669763e-07,
+      "loss": 0.2844,
+      "step": 25083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.001704186742299e-07,
+      "loss": 0.3197,
+      "step": 25084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.997057755758554e-07,
+      "loss": 0.3185,
+      "step": 25085
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.992412467776868e-07,
+      "loss": 0.3516,
+      "step": 25086
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.98776832285555e-07,
+      "loss": 0.2875,
+      "step": 25087
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.983125321052955e-07,
+      "loss": 0.3192,
+      "step": 25088
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.978483462427423e-07,
+      "loss": 0.3422,
+      "step": 25089
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.973842747037221e-07,
+      "loss": 0.2972,
+      "step": 25090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.969203174940655e-07,
+      "loss": 0.319,
+      "step": 25091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.964564746196014e-07,
+      "loss": 0.3339,
+      "step": 25092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.959927460861517e-07,
+      "loss": 0.2786,
+      "step": 25093
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.955291318995429e-07,
+      "loss": 0.3348,
+      "step": 25094
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.950656320656003e-07,
+      "loss": 0.3027,
+      "step": 25095
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.946022465901428e-07,
+      "loss": 0.3309,
+      "step": 25096
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.941389754789898e-07,
+      "loss": 0.3218,
+      "step": 25097
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.936758187379624e-07,
+      "loss": 0.3111,
+      "step": 25098
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.932127763728759e-07,
+      "loss": 0.3088,
+      "step": 25099
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.927498483895458e-07,
+      "loss": 0.3146,
+      "step": 25100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.922870347937907e-07,
+      "loss": 0.3192,
+      "step": 25101
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.91824335591418e-07,
+      "loss": 0.3072,
+      "step": 25102
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.91361750788241e-07,
+      "loss": 0.282,
+      "step": 25103
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.908992803900729e-07,
+      "loss": 0.2647,
+      "step": 25104
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.904369244027189e-07,
+      "loss": 0.3314,
+      "step": 25105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.899746828319888e-07,
+      "loss": 0.3212,
+      "step": 25106
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.89512555683687e-07,
+      "loss": 0.3117,
+      "step": 25107
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.890505429636153e-07,
+      "loss": 0.3221,
+      "step": 25108
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.885886446775793e-07,
+      "loss": 0.356,
+      "step": 25109
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.881268608313831e-07,
+      "loss": 0.2873,
+      "step": 25110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.87665191430821e-07,
+      "loss": 0.3025,
+      "step": 25111
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.872036364816938e-07,
+      "loss": 0.3132,
+      "step": 25112
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.867421959898026e-07,
+      "loss": 0.3122,
+      "step": 25113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.862808699609371e-07,
+      "loss": 0.3437,
+      "step": 25114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.858196584008949e-07,
+      "loss": 0.2769,
+      "step": 25115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.853585613154691e-07,
+      "loss": 0.3199,
+      "step": 25116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.848975787104508e-07,
+      "loss": 0.2618,
+      "step": 25117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.844367105916273e-07,
+      "loss": 0.2971,
+      "step": 25118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.839759569647921e-07,
+      "loss": 0.3039,
+      "step": 25119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.835153178357269e-07,
+      "loss": 0.3192,
+      "step": 25120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.830547932102196e-07,
+      "loss": 0.2797,
+      "step": 25121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.825943830940554e-07,
+      "loss": 0.3189,
+      "step": 25122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.821340874930173e-07,
+      "loss": 0.3121,
+      "step": 25123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.816739064128843e-07,
+      "loss": 0.2813,
+      "step": 25124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.812138398594383e-07,
+      "loss": 0.2798,
+      "step": 25125
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.80753887838458e-07,
+      "loss": 0.3238,
+      "step": 25126
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.8029405035572e-07,
+      "loss": 0.3302,
+      "step": 25127
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.798343274169973e-07,
+      "loss": 0.2982,
+      "step": 25128
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.793747190280689e-07,
+      "loss": 0.3041,
+      "step": 25129
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.78915225194703e-07,
+      "loss": 0.2916,
+      "step": 25130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.784558459226722e-07,
+      "loss": 0.3172,
+      "step": 25131
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.779965812177482e-07,
+      "loss": 0.2973,
+      "step": 25132
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.775374310856977e-07,
+      "loss": 0.3209,
+      "step": 25133
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.770783955322871e-07,
+      "loss": 0.2857,
+      "step": 25134
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.766194745632839e-07,
+      "loss": 0.2991,
+      "step": 25135
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.761606681844503e-07,
+      "loss": 0.3312,
+      "step": 25136
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.757019764015506e-07,
+      "loss": 0.3128,
+      "step": 25137
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.752433992203458e-07,
+      "loss": 0.3197,
+      "step": 25138
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.747849366465933e-07,
+      "loss": 0.3241,
+      "step": 25139
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.74326588686053e-07,
+      "loss": 0.2931,
+      "step": 25140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.738683553444837e-07,
+      "loss": 0.3116,
+      "step": 25141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.734102366276364e-07,
+      "loss": 0.3393,
+      "step": 25142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.729522325412688e-07,
+      "loss": 0.3019,
+      "step": 25143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.724943430911325e-07,
+      "loss": 0.2991,
+      "step": 25144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.720365682829779e-07,
+      "loss": 0.3053,
+      "step": 25145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.715789081225556e-07,
+      "loss": 0.3257,
+      "step": 25146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.711213626156145e-07,
+      "loss": 0.328,
+      "step": 25147
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.706639317679e-07,
+      "loss": 0.3034,
+      "step": 25148
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.702066155851563e-07,
+      "loss": 0.2845,
+      "step": 25149
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.69749414073131e-07,
+      "loss": 0.2372,
+      "step": 25150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.692923272375631e-07,
+      "loss": 0.3056,
+      "step": 25151
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.688353550841944e-07,
+      "loss": 0.3234,
+      "step": 25152
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.683784976187659e-07,
+      "loss": 0.3255,
+      "step": 25153
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.679217548470142e-07,
+      "loss": 0.3548,
+      "step": 25154
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.674651267746769e-07,
+      "loss": 0.2942,
+      "step": 25155
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.670086134074896e-07,
+      "loss": 0.3205,
+      "step": 25156
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.665522147511851e-07,
+      "loss": 0.3288,
+      "step": 25157
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.66095930811498e-07,
+      "loss": 0.3134,
+      "step": 25158
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.656397615941547e-07,
+      "loss": 0.2883,
+      "step": 25159
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.651837071048907e-07,
+      "loss": 0.3379,
+      "step": 25160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.647277673494281e-07,
+      "loss": 0.3472,
+      "step": 25161
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.642719423334967e-07,
+      "loss": 0.2681,
+      "step": 25162
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.63816232062824e-07,
+      "loss": 0.3022,
+      "step": 25163
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.63360636543128e-07,
+      "loss": 0.2755,
+      "step": 25164
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.629051557801349e-07,
+      "loss": 0.31,
+      "step": 25165
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.62449789779567e-07,
+      "loss": 0.3623,
+      "step": 25166
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.619945385471384e-07,
+      "loss": 0.2935,
+      "step": 25167
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.615394020885726e-07,
+      "loss": 0.3079,
+      "step": 25168
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.610843804095826e-07,
+      "loss": 0.2809,
+      "step": 25169
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.606294735158849e-07,
+      "loss": 0.3132,
+      "step": 25170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.601746814131928e-07,
+      "loss": 0.2848,
+      "step": 25171
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.597200041072196e-07,
+      "loss": 0.3389,
+      "step": 25172
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.592654416036727e-07,
+      "loss": 0.2979,
+      "step": 25173
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.588109939082645e-07,
+      "loss": 0.3221,
+      "step": 25174
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.583566610267036e-07,
+      "loss": 0.303,
+      "step": 25175
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.579024429646932e-07,
+      "loss": 0.3448,
+      "step": 25176
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.574483397279409e-07,
+      "loss": 0.3209,
+      "step": 25177
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.569943513221502e-07,
+      "loss": 0.3022,
+      "step": 25178
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.565404777530218e-07,
+      "loss": 0.2778,
+      "step": 25179
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.560867190262579e-07,
+      "loss": 0.3059,
+      "step": 25180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.556330751475572e-07,
+      "loss": 0.3321,
+      "step": 25181
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.551795461226165e-07,
+      "loss": 0.3103,
+      "step": 25182
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.547261319571321e-07,
+      "loss": 0.3478,
+      "step": 25183
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.542728326568006e-07,
+      "loss": 0.2955,
+      "step": 25184
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.538196482273142e-07,
+      "loss": 0.3123,
+      "step": 25185
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.533665786743639e-07,
+      "loss": 0.3138,
+      "step": 25186
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.529136240036439e-07,
+      "loss": 0.3262,
+      "step": 25187
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.524607842208398e-07,
+      "loss": 0.3341,
+      "step": 25188
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.520080593316405e-07,
+      "loss": 0.3472,
+      "step": 25189
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.515554493417333e-07,
+      "loss": 0.3113,
+      "step": 25190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.511029542568028e-07,
+      "loss": 0.3026,
+      "step": 25191
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.506505740825299e-07,
+      "loss": 0.2978,
+      "step": 25192
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.501983088245991e-07,
+      "loss": 0.2894,
+      "step": 25193
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.49746158488689e-07,
+      "loss": 0.2723,
+      "step": 25194
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.492941230804797e-07,
+      "loss": 0.278,
+      "step": 25195
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.488422026056487e-07,
+      "loss": 0.3198,
+      "step": 25196
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.483903970698737e-07,
+      "loss": 0.2709,
+      "step": 25197
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.479387064788258e-07,
+      "loss": 0.287,
+      "step": 25198
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.474871308381804e-07,
+      "loss": 0.3086,
+      "step": 25199
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.470356701536109e-07,
+      "loss": 0.3535,
+      "step": 25200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.465843244307859e-07,
+      "loss": 0.2891,
+      "step": 25201
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.461330936753731e-07,
+      "loss": 0.3162,
+      "step": 25202
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.456819778930436e-07,
+      "loss": 0.2699,
+      "step": 25203
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.452309770894584e-07,
+      "loss": 0.3076,
+      "step": 25204
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.447800912702853e-07,
+      "loss": 0.3258,
+      "step": 25205
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.443293204411884e-07,
+      "loss": 0.3188,
+      "step": 25206
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.438786646078267e-07,
+      "loss": 0.2978,
+      "step": 25207
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.434281237758624e-07,
+      "loss": 0.3302,
+      "step": 25208
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.42977697950954e-07,
+      "loss": 0.3279,
+      "step": 25209
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.425273871387573e-07,
+      "loss": 0.3248,
+      "step": 25210
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.420771913449322e-07,
+      "loss": 0.3119,
+      "step": 25211
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.416271105751294e-07,
+      "loss": 0.3068,
+      "step": 25212
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.411771448350015e-07,
+      "loss": 0.3284,
+      "step": 25213
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.407272941302014e-07,
+      "loss": 0.3012,
+      "step": 25214
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.402775584663814e-07,
+      "loss": 0.2831,
+      "step": 25215
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.39827937849187e-07,
+      "loss": 0.2868,
+      "step": 25216
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.393784322842669e-07,
+      "loss": 0.2983,
+      "step": 25217
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.389290417772677e-07,
+      "loss": 0.3221,
+      "step": 25218
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.384797663338307e-07,
+      "loss": 0.3038,
+      "step": 25219
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.380306059596022e-07,
+      "loss": 0.3095,
+      "step": 25220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.375815606602234e-07,
+      "loss": 0.3239,
+      "step": 25221
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.37132630441333e-07,
+      "loss": 0.3372,
+      "step": 25222
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.366838153085688e-07,
+      "loss": 0.3269,
+      "step": 25223
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.362351152675708e-07,
+      "loss": 0.3029,
+      "step": 25224
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.35786530323971e-07,
+      "loss": 0.3052,
+      "step": 25225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.35338060483406e-07,
+      "loss": 0.3048,
+      "step": 25226
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.348897057515093e-07,
+      "loss": 0.2925,
+      "step": 25227
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.344414661339106e-07,
+      "loss": 0.3233,
+      "step": 25228
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.339933416362389e-07,
+      "loss": 0.3092,
+      "step": 25229
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.335453322641274e-07,
+      "loss": 0.3253,
+      "step": 25230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.330974380231971e-07,
+      "loss": 0.2867,
+      "step": 25231
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.326496589190791e-07,
+      "loss": 0.3039,
+      "step": 25232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.322019949573923e-07,
+      "loss": 0.32,
+      "step": 25233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.317544461437633e-07,
+      "loss": 0.2914,
+      "step": 25234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.313070124838107e-07,
+      "loss": 0.3023,
+      "step": 25235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.308596939831559e-07,
+      "loss": 0.3053,
+      "step": 25236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.304124906474186e-07,
+      "loss": 0.3155,
+      "step": 25237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.299654024822113e-07,
+      "loss": 0.2985,
+      "step": 25238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.295184294931524e-07,
+      "loss": 0.2891,
+      "step": 25239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.290715716858578e-07,
+      "loss": 0.3134,
+      "step": 25240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.28624829065936e-07,
+      "loss": 0.2902,
+      "step": 25241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.281782016390028e-07,
+      "loss": 0.3061,
+      "step": 25242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.277316894106613e-07,
+      "loss": 0.3016,
+      "step": 25243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.272852923865271e-07,
+      "loss": 0.3171,
+      "step": 25244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.268390105722002e-07,
+      "loss": 0.345,
+      "step": 25245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.263928439732916e-07,
+      "loss": 0.3029,
+      "step": 25246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.259467925954001e-07,
+      "loss": 0.303,
+      "step": 25247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.255008564441314e-07,
+      "loss": 0.3361,
+      "step": 25248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250550355250875e-07,
+      "loss": 0.3047,
+      "step": 25249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.246093298438651e-07,
+      "loss": 0.3025,
+      "step": 25250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.241637394060619e-07,
+      "loss": 0.302,
+      "step": 25251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.237182642172792e-07,
+      "loss": 0.315,
+      "step": 25252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.232729042831067e-07,
+      "loss": 0.2922,
+      "step": 25253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.228276596091411e-07,
+      "loss": 0.3011,
+      "step": 25254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.223825302009758e-07,
+      "loss": 0.2949,
+      "step": 25255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.219375160641973e-07,
+      "loss": 0.3057,
+      "step": 25256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.214926172043979e-07,
+      "loss": 0.3052,
+      "step": 25257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.210478336271665e-07,
+      "loss": 0.3246,
+      "step": 25258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.206031653380875e-07,
+      "loss": 0.3312,
+      "step": 25259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.201586123427463e-07,
+      "loss": 0.3008,
+      "step": 25260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.197141746467297e-07,
+      "loss": 0.324,
+      "step": 25261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.192698522556142e-07,
+      "loss": 0.285,
+      "step": 25262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.188256451749843e-07,
+      "loss": 0.2987,
+      "step": 25263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.1838155341042e-07,
+      "loss": 0.3015,
+      "step": 25264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.179375769674969e-07,
+      "loss": 0.3289,
+      "step": 25265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.174937158517914e-07,
+      "loss": 0.2949,
+      "step": 25266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.170499700688806e-07,
+      "loss": 0.3282,
+      "step": 25267
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.166063396243351e-07,
+      "loss": 0.2878,
+      "step": 25268
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.161628245237274e-07,
+      "loss": 0.2819,
+      "step": 25269
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.157194247726297e-07,
+      "loss": 0.2922,
+      "step": 25270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.152761403766118e-07,
+      "loss": 0.3085,
+      "step": 25271
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.148329713412384e-07,
+      "loss": 0.3152,
+      "step": 25272
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.143899176720782e-07,
+      "loss": 0.3108,
+      "step": 25273
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.139469793746946e-07,
+      "loss": 0.3326,
+      "step": 25274
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.13504156454652e-07,
+      "loss": 0.3101,
+      "step": 25275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.130614489175115e-07,
+      "loss": 0.3121,
+      "step": 25276
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.126188567688342e-07,
+      "loss": 0.3251,
+      "step": 25277
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.121763800141769e-07,
+      "loss": 0.2967,
+      "step": 25278
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.117340186590994e-07,
+      "loss": 0.3093,
+      "step": 25279
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.112917727091585e-07,
+      "loss": 0.3045,
+      "step": 25280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.108496421699063e-07,
+      "loss": 0.3361,
+      "step": 25281
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.104076270468975e-07,
+      "loss": 0.3097,
+      "step": 25282
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.09965727345684e-07,
+      "loss": 0.3119,
+      "step": 25283
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.095239430718149e-07,
+      "loss": 0.3104,
+      "step": 25284
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.090822742308413e-07,
+      "loss": 0.2899,
+      "step": 25285
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.086407208283098e-07,
+      "loss": 0.283,
+      "step": 25286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.081992828697626e-07,
+      "loss": 0.3066,
+      "step": 25287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.077579603607477e-07,
+      "loss": 0.3152,
+      "step": 25288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.073167533068105e-07,
+      "loss": 0.323,
+      "step": 25289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.068756617134865e-07,
+      "loss": 0.3144,
+      "step": 25290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.064346855863192e-07,
+      "loss": 0.3204,
+      "step": 25291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.059938249308496e-07,
+      "loss": 0.3087,
+      "step": 25292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.055530797526101e-07,
+      "loss": 0.2934,
+      "step": 25293
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.051124500571394e-07,
+      "loss": 0.313,
+      "step": 25294
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.046719358499733e-07,
+      "loss": 0.3325,
+      "step": 25295
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.042315371366416e-07,
+      "loss": 0.3216,
+      "step": 25296
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.037912539226755e-07,
+      "loss": 0.2386,
+      "step": 25297
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.033510862136085e-07,
+      "loss": 0.3279,
+      "step": 25298
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.029110340149648e-07,
+      "loss": 0.2801,
+      "step": 25299
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.024710973322736e-07,
+      "loss": 0.3019,
+      "step": 25300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.020312761710625e-07,
+      "loss": 0.2923,
+      "step": 25301
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.015915705368527e-07,
+      "loss": 0.3167,
+      "step": 25302
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.011519804351664e-07,
+      "loss": 0.2873,
+      "step": 25303
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.007125058715271e-07,
+      "loss": 0.3151,
+      "step": 25304
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.002731468514557e-07,
+      "loss": 0.3217,
+      "step": 25305
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.998339033804692e-07,
+      "loss": 0.3056,
+      "step": 25306
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.993947754640819e-07,
+      "loss": 0.3048,
+      "step": 25307
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.989557631078138e-07,
+      "loss": 0.2922,
+      "step": 25308
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.985168663171739e-07,
+      "loss": 0.3287,
+      "step": 25309
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.980780850976777e-07,
+      "loss": 0.3183,
+      "step": 25310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.976394194548387e-07,
+      "loss": 0.2871,
+      "step": 25311
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.972008693941613e-07,
+      "loss": 0.3033,
+      "step": 25312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.967624349211578e-07,
+      "loss": 0.3253,
+      "step": 25313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.963241160413338e-07,
+      "loss": 0.3082,
+      "step": 25314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.958859127601937e-07,
+      "loss": 0.2915,
+      "step": 25315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.954478250832431e-07,
+      "loss": 0.287,
+      "step": 25316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.950098530159834e-07,
+      "loss": 0.3028,
+      "step": 25317
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.945719965639132e-07,
+      "loss": 0.3008,
+      "step": 25318
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.941342557325349e-07,
+      "loss": 0.2958,
+      "step": 25319
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.936966305273486e-07,
+      "loss": 0.3306,
+      "step": 25320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.932591209538454e-07,
+      "loss": 0.2632,
+      "step": 25321
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.928217270175231e-07,
+      "loss": 0.2924,
+      "step": 25322
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.923844487238775e-07,
+      "loss": 0.3307,
+      "step": 25323
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.919472860783972e-07,
+      "loss": 0.3136,
+      "step": 25324
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.915102390865748e-07,
+      "loss": 0.3319,
+      "step": 25325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.910733077539012e-07,
+      "loss": 0.2952,
+      "step": 25326
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.906364920858633e-07,
+      "loss": 0.3112,
+      "step": 25327
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.901997920879445e-07,
+      "loss": 0.2663,
+      "step": 25328
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.897632077656336e-07,
+      "loss": 0.3525,
+      "step": 25329
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.893267391244108e-07,
+      "loss": 0.2899,
+      "step": 25330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.888903861697606e-07,
+      "loss": 0.3007,
+      "step": 25331
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.884541489071651e-07,
+      "loss": 0.3285,
+      "step": 25332
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.88018027342099e-07,
+      "loss": 0.2998,
+      "step": 25333
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.875820214800423e-07,
+      "loss": 0.312,
+      "step": 25334
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.871461313264728e-07,
+      "loss": 0.2957,
+      "step": 25335
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.867103568868628e-07,
+      "loss": 0.3261,
+      "step": 25336
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.86274698166688e-07,
+      "loss": 0.2665,
+      "step": 25337
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.858391551714184e-07,
+      "loss": 0.2815,
+      "step": 25338
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.854037279065252e-07,
+      "loss": 0.3251,
+      "step": 25339
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.849684163774762e-07,
+      "loss": 0.3147,
+      "step": 25340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.845332205897404e-07,
+      "loss": 0.3183,
+      "step": 25341
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.840981405487847e-07,
+      "loss": 0.3187,
+      "step": 25342
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.83663176260071e-07,
+      "loss": 0.3215,
+      "step": 25343
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.832283277290643e-07,
+      "loss": 0.3332,
+      "step": 25344
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.827935949612275e-07,
+      "loss": 0.3093,
+      "step": 25345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.823589779620167e-07,
+      "loss": 0.3281,
+      "step": 25346
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.81924476736896e-07,
+      "loss": 0.2462,
+      "step": 25347
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.814900912913181e-07,
+      "loss": 0.3227,
+      "step": 25348
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.81055821630744e-07,
+      "loss": 0.3186,
+      "step": 25349
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.806216677606215e-07,
+      "loss": 0.2846,
+      "step": 25350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.801876296864097e-07,
+      "loss": 0.3525,
+      "step": 25351
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.797537074135564e-07,
+      "loss": 0.3698,
+      "step": 25352
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.793199009475117e-07,
+      "loss": 0.3328,
+      "step": 25353
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.788862102937289e-07,
+      "loss": 0.3288,
+      "step": 25354
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.784526354576483e-07,
+      "loss": 0.3273,
+      "step": 25355
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.780191764447198e-07,
+      "loss": 0.3049,
+      "step": 25356
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.775858332603892e-07,
+      "loss": 0.2971,
+      "step": 25357
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.771526059100942e-07,
+      "loss": 0.3162,
+      "step": 25358
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.767194943992817e-07,
+      "loss": 0.2822,
+      "step": 25359
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.762864987333885e-07,
+      "loss": 0.3164,
+      "step": 25360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.758536189178512e-07,
+      "loss": 0.3044,
+      "step": 25361
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.7542085495811e-07,
+      "loss": 0.296,
+      "step": 25362
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.749882068596004e-07,
+      "loss": 0.3016,
+      "step": 25363
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.74555674627755e-07,
+      "loss": 0.3105,
+      "step": 25364
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.741232582680058e-07,
+      "loss": 0.3233,
+      "step": 25365
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.736909577857876e-07,
+      "loss": 0.2705,
+      "step": 25366
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.73258773186526e-07,
+      "loss": 0.3188,
+      "step": 25367
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.72826704475651e-07,
+      "loss": 0.2731,
+      "step": 25368
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.723947516585906e-07,
+      "loss": 0.2946,
+      "step": 25369
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.719629147407692e-07,
+      "loss": 0.2879,
+      "step": 25370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.715311937276082e-07,
+      "loss": 0.284,
+      "step": 25371
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.710995886245342e-07,
+      "loss": 0.3199,
+      "step": 25372
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.706680994369653e-07,
+      "loss": 0.3077,
+      "step": 25373
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.702367261703215e-07,
+      "loss": 0.3025,
+      "step": 25374
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.698054688300227e-07,
+      "loss": 0.3137,
+      "step": 25375
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.693743274214827e-07,
+      "loss": 0.2941,
+      "step": 25376
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.68943301950118e-07,
+      "loss": 0.3401,
+      "step": 25377
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.685123924213422e-07,
+      "loss": 0.2903,
+      "step": 25378
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.680815988405698e-07,
+      "loss": 0.2365,
+      "step": 25379
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.6765092121321e-07,
+      "loss": 0.3246,
+      "step": 25380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.672203595446703e-07,
+      "loss": 0.3167,
+      "step": 25381
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.667899138403601e-07,
+      "loss": 0.2854,
+      "step": 25382
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.663595841056847e-07,
+      "loss": 0.2855,
+      "step": 25383
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.659293703460513e-07,
+      "loss": 0.275,
+      "step": 25384
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.65499272566862e-07,
+      "loss": 0.3317,
+      "step": 25385
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.650692907735191e-07,
+      "loss": 0.3183,
+      "step": 25386
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.646394249714218e-07,
+      "loss": 0.2993,
+      "step": 25387
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.642096751659723e-07,
+      "loss": 0.3095,
+      "step": 25388
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.637800413625652e-07,
+      "loss": 0.3072,
+      "step": 25389
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.633505235665995e-07,
+      "loss": 0.2902,
+      "step": 25390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.629211217834686e-07,
+      "loss": 0.3092,
+      "step": 25391
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.624918360185651e-07,
+      "loss": 0.2754,
+      "step": 25392
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.620626662772801e-07,
+      "loss": 0.3025,
+      "step": 25393
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.616336125650081e-07,
+      "loss": 0.3316,
+      "step": 25394
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.3143,
+      "step": 25395
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.60775853249045e-07,
+      "loss": 0.298,
+      "step": 25396
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.603471476561309e-07,
+      "loss": 0.2821,
+      "step": 25397
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.599185581137725e-07,
+      "loss": 0.3058,
+      "step": 25398
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.594900846273545e-07,
+      "loss": 0.3376,
+      "step": 25399
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.590617272022605e-07,
+      "loss": 0.3181,
+      "step": 25400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.586334858438693e-07,
+      "loss": 0.2964,
+      "step": 25401
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.582053605575568e-07,
+      "loss": 0.3333,
+      "step": 25402
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.577773513487042e-07,
+      "loss": 0.3232,
+      "step": 25403
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.573494582226837e-07,
+      "loss": 0.2845,
+      "step": 25404
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.569216811848723e-07,
+      "loss": 0.3241,
+      "step": 25405
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.564940202406434e-07,
+      "loss": 0.3456,
+      "step": 25406
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.560664753953662e-07,
+      "loss": 0.3155,
+      "step": 25407
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.556390466544117e-07,
+      "loss": 0.2972,
+      "step": 25408
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.552117340231501e-07,
+      "loss": 0.3071,
+      "step": 25409
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.547845375069452e-07,
+      "loss": 0.3123,
+      "step": 25410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.543574571111656e-07,
+      "loss": 0.3186,
+      "step": 25411
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.539304928411739e-07,
+      "loss": 0.2942,
+      "step": 25412
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.535036447023325e-07,
+      "loss": 0.3574,
+      "step": 25413
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.530769127000026e-07,
+      "loss": 0.3368,
+      "step": 25414
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.526502968395444e-07,
+      "loss": 0.316,
+      "step": 25415
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.52223797126318e-07,
+      "loss": 0.2944,
+      "step": 25416
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.51797413565677e-07,
+      "loss": 0.334,
+      "step": 25417
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.513711461629792e-07,
+      "loss": 0.3351,
+      "step": 25418
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.509449949235781e-07,
+      "loss": 0.3498,
+      "step": 25419
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.505189598528251e-07,
+      "loss": 0.3552,
+      "step": 25420
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.500930409560736e-07,
+      "loss": 0.3153,
+      "step": 25421
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.496672382386716e-07,
+      "loss": 0.3104,
+      "step": 25422
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.492415517059659e-07,
+      "loss": 0.3212,
+      "step": 25423
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.488159813633033e-07,
+      "loss": 0.3182,
+      "step": 25424
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.48390527216033e-07,
+      "loss": 0.325,
+      "step": 25425
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.479651892694928e-07,
+      "loss": 0.321,
+      "step": 25426
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.475399675290285e-07,
+      "loss": 0.3328,
+      "step": 25427
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.471148619999824e-07,
+      "loss": 0.2747,
+      "step": 25428
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.466898726876904e-07,
+      "loss": 0.2634,
+      "step": 25429
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.462649995974914e-07,
+      "loss": 0.3126,
+      "step": 25430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.458402427347244e-07,
+      "loss": 0.339,
+      "step": 25431
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.454156021047221e-07,
+      "loss": 0.3137,
+      "step": 25432
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.449910777128155e-07,
+      "loss": 0.2991,
+      "step": 25433
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.445666695643417e-07,
+      "loss": 0.3044,
+      "step": 25434
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.441423776646272e-07,
+      "loss": 0.3245,
+      "step": 25435
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.437182020190036e-07,
+      "loss": 0.2977,
+      "step": 25436
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.432941426327989e-07,
+      "loss": 0.279,
+      "step": 25437
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.428701995113363e-07,
+      "loss": 0.2902,
+      "step": 25438
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.424463726599417e-07,
+      "loss": 0.2591,
+      "step": 25439
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.420226620839421e-07,
+      "loss": 0.3089,
+      "step": 25440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.415990677886542e-07,
+      "loss": 0.3491,
+      "step": 25441
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.411755897794026e-07,
+      "loss": 0.3183,
+      "step": 25442
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.40752228061502e-07,
+      "loss": 0.3155,
+      "step": 25443
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.403289826402737e-07,
+      "loss": 0.3114,
+      "step": 25444
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.399058535210301e-07,
+      "loss": 0.331,
+      "step": 25445
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394828407090893e-07,
+      "loss": 0.344,
+      "step": 25446
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.390599442097613e-07,
+      "loss": 0.308,
+      "step": 25447
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.386371640283597e-07,
+      "loss": 0.3021,
+      "step": 25448
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.382145001701946e-07,
+      "loss": 0.3125,
+      "step": 25449
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.37791952640573e-07,
+      "loss": 0.2806,
+      "step": 25450
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.373695214448029e-07,
+      "loss": 0.3312,
+      "step": 25451
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.369472065881922e-07,
+      "loss": 0.2872,
+      "step": 25452
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.365250080760422e-07,
+      "loss": 0.2758,
+      "step": 25453
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.361029259136577e-07,
+      "loss": 0.2788,
+      "step": 25454
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.356809601063386e-07,
+      "loss": 0.2631,
+      "step": 25455
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.352591106593864e-07,
+      "loss": 0.3459,
+      "step": 25456
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.34837377578097e-07,
+      "loss": 0.3264,
+      "step": 25457
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.344157608677694e-07,
+      "loss": 0.3268,
+      "step": 25458
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.339942605337002e-07,
+      "loss": 0.3145,
+      "step": 25459
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.3357287658118e-07,
+      "loss": 0.329,
+      "step": 25460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.331516090155044e-07,
+      "loss": 0.2973,
+      "step": 25461
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.327304578419648e-07,
+      "loss": 0.3219,
+      "step": 25462
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.323094230658479e-07,
+      "loss": 0.2878,
+      "step": 25463
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.318885046924451e-07,
+      "loss": 0.3021,
+      "step": 25464
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.314677027270423e-07,
+      "loss": 0.3479,
+      "step": 25465
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.310470171749218e-07,
+      "loss": 0.3003,
+      "step": 25466
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.306264480413705e-07,
+      "loss": 0.3223,
+      "step": 25467
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.30205995331672e-07,
+      "loss": 0.2869,
+      "step": 25468
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.297856590511032e-07,
+      "loss": 0.2942,
+      "step": 25469
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.293654392049466e-07,
+      "loss": 0.3059,
+      "step": 25470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.289453357984799e-07,
+      "loss": 0.2758,
+      "step": 25471
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.28525348836977e-07,
+      "loss": 0.3001,
+      "step": 25472
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.281054783257158e-07,
+      "loss": 0.3308,
+      "step": 25473
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.276857242699708e-07,
+      "loss": 0.2981,
+      "step": 25474
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.272660866750125e-07,
+      "loss": 0.2783,
+      "step": 25475
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.268465655461099e-07,
+      "loss": 0.2916,
+      "step": 25476
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.264271608885353e-07,
+      "loss": 0.3109,
+      "step": 25477
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.260078727075525e-07,
+      "loss": 0.339,
+      "step": 25478
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.255887010084317e-07,
+      "loss": 0.2736,
+      "step": 25479
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.251696457964363e-07,
+      "loss": 0.3352,
+      "step": 25480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.247507070768289e-07,
+      "loss": 0.3196,
+      "step": 25481
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.24331884854872e-07,
+      "loss": 0.3042,
+      "step": 25482
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.239131791358278e-07,
+      "loss": 0.3024,
+      "step": 25483
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.234945899249513e-07,
+      "loss": 0.311,
+      "step": 25484
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.230761172275047e-07,
+      "loss": 0.3335,
+      "step": 25485
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.226577610487417e-07,
+      "loss": 0.3127,
+      "step": 25486
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.222395213939148e-07,
+      "loss": 0.3449,
+      "step": 25487
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.218213982682787e-07,
+      "loss": 0.3024,
+      "step": 25488
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.214033916770868e-07,
+      "loss": 0.3088,
+      "step": 25489
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.209855016255884e-07,
+      "loss": 0.2958,
+      "step": 25490
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.205677281190304e-07,
+      "loss": 0.3057,
+      "step": 25491
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.201500711626619e-07,
+      "loss": 0.2888,
+      "step": 25492
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.197325307617287e-07,
+      "loss": 0.3212,
+      "step": 25493
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.193151069214743e-07,
+      "loss": 0.2964,
+      "step": 25494
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.188977996471435e-07,
+      "loss": 0.3336,
+      "step": 25495
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.184806089439755e-07,
+      "loss": 0.2974,
+      "step": 25496
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.180635348172094e-07,
+      "loss": 0.2704,
+      "step": 25497
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.176465772720853e-07,
+      "loss": 0.3223,
+      "step": 25498
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.172297363138414e-07,
+      "loss": 0.3029,
+      "step": 25499
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.168130119477102e-07,
+      "loss": 0.3107,
+      "step": 25500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.163964041789284e-07,
+      "loss": 0.2962,
+      "step": 25501
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.159799130127288e-07,
+      "loss": 0.336,
+      "step": 25502
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.155635384543391e-07,
+      "loss": 0.3216,
+      "step": 25503
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.151472805089921e-07,
+      "loss": 0.3149,
+      "step": 25504
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.147311391819167e-07,
+      "loss": 0.2864,
+      "step": 25505
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.14315114478339e-07,
+      "loss": 0.3042,
+      "step": 25506
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.138992064034811e-07,
+      "loss": 0.2698,
+      "step": 25507
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.134834149625703e-07,
+      "loss": 0.2928,
+      "step": 25508
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.130677401608277e-07,
+      "loss": 0.3017,
+      "step": 25509
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.126521820034737e-07,
+      "loss": 0.2834,
+      "step": 25510
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.122367404957297e-07,
+      "loss": 0.3135,
+      "step": 25511
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.118214156428105e-07,
+      "loss": 0.2938,
+      "step": 25512
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.114062074499361e-07,
+      "loss": 0.3067,
+      "step": 25513
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.109911159223204e-07,
+      "loss": 0.2723,
+      "step": 25514
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.105761410651746e-07,
+      "loss": 0.2899,
+      "step": 25515
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.101612828837157e-07,
+      "loss": 0.2858,
+      "step": 25516
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.097465413831484e-07,
+      "loss": 0.3083,
+      "step": 25517
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.093319165686874e-07,
+      "loss": 0.2971,
+      "step": 25518
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.089174084455352e-07,
+      "loss": 0.3175,
+      "step": 25519
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.085030170189034e-07,
+      "loss": 0.3247,
+      "step": 25520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08088742293992e-07,
+      "loss": 0.2996,
+      "step": 25521
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.076745842760057e-07,
+      "loss": 0.2588,
+      "step": 25522
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.072605429701484e-07,
+      "loss": 0.3402,
+      "step": 25523
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.068466183816203e-07,
+      "loss": 0.3275,
+      "step": 25524
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.06432810515617e-07,
+      "loss": 0.3232,
+      "step": 25525
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.060191193773403e-07,
+      "loss": 0.3634,
+      "step": 25526
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.056055449719834e-07,
+      "loss": 0.2876,
+      "step": 25527
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.051920873047403e-07,
+      "loss": 0.309,
+      "step": 25528
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.047787463808054e-07,
+      "loss": 0.272,
+      "step": 25529
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.043655222053714e-07,
+      "loss": 0.2924,
+      "step": 25530
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.039524147836251e-07,
+      "loss": 0.2811,
+      "step": 25531
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.035394241207571e-07,
+      "loss": 0.3411,
+      "step": 25532
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.031265502219565e-07,
+      "loss": 0.3153,
+      "step": 25533
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.027137930924055e-07,
+      "loss": 0.2862,
+      "step": 25534
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.023011527372903e-07,
+      "loss": 0.3169,
+      "step": 25535
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.018886291617944e-07,
+      "loss": 0.3334,
+      "step": 25536
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.014762223710981e-07,
+      "loss": 0.3242,
+      "step": 25537
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.010639323703794e-07,
+      "loss": 0.2549,
+      "step": 25538
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.00651759164821e-07,
+      "loss": 0.3442,
+      "step": 25539
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.002397027595942e-07,
+      "loss": 0.3271,
+      "step": 25540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.998277631598793e-07,
+      "loss": 0.3096,
+      "step": 25541
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.99415940370849e-07,
+      "loss": 0.3184,
+      "step": 25542
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.990042343976744e-07,
+      "loss": 0.3032,
+      "step": 25543
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.985926452455272e-07,
+      "loss": 0.3069,
+      "step": 25544
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.981811729195786e-07,
+      "loss": 0.33,
+      "step": 25545
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.977698174249948e-07,
+      "loss": 0.2891,
+      "step": 25546
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.973585787669434e-07,
+      "loss": 0.2939,
+      "step": 25547
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.969474569505885e-07,
+      "loss": 0.3018,
+      "step": 25548
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.965364519810969e-07,
+      "loss": 0.2841,
+      "step": 25549
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.961255638636256e-07,
+      "loss": 0.3219,
+      "step": 25550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.957147926033403e-07,
+      "loss": 0.2653,
+      "step": 25551
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.953041382053971e-07,
+      "loss": 0.3641,
+      "step": 25552
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.948936006749541e-07,
+      "loss": 0.3146,
+      "step": 25553
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.944831800171714e-07,
+      "loss": 0.2638,
+      "step": 25554
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.940728762371985e-07,
+      "loss": 0.3,
+      "step": 25555
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.936626893401921e-07,
+      "loss": 0.2891,
+      "step": 25556
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.932526193313049e-07,
+      "loss": 0.3366,
+      "step": 25557
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.928426662156851e-07,
+      "loss": 0.3085,
+      "step": 25558
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924328299984829e-07,
+      "loss": 0.3248,
+      "step": 25559
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.920231106848452e-07,
+      "loss": 0.2842,
+      "step": 25560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.916135082799191e-07,
+      "loss": 0.3523,
+      "step": 25561
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.912040227888484e-07,
+      "loss": 0.3281,
+      "step": 25562
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.907946542167754e-07,
+      "loss": 0.305,
+      "step": 25563
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.90385402568845e-07,
+      "loss": 0.2907,
+      "step": 25564
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.899762678501942e-07,
+      "loss": 0.3514,
+      "step": 25565
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.895672500659623e-07,
+      "loss": 0.3198,
+      "step": 25566
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.891583492212883e-07,
+      "loss": 0.3016,
+      "step": 25567
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.887495653213062e-07,
+      "loss": 0.2983,
+      "step": 25568
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.883408983711525e-07,
+      "loss": 0.3139,
+      "step": 25569
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.879323483759582e-07,
+      "loss": 0.2927,
+      "step": 25570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.875239153408541e-07,
+      "loss": 0.3083,
+      "step": 25571
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.8711559927097e-07,
+      "loss": 0.302,
+      "step": 25572
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.86707400171438e-07,
+      "loss": 0.3028,
+      "step": 25573
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.86299318047381e-07,
+      "loss": 0.3011,
+      "step": 25574
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.858913529039268e-07,
+      "loss": 0.3157,
+      "step": 25575
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.854835047461994e-07,
+      "loss": 0.3255,
+      "step": 25576
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.8507577357932e-07,
+      "loss": 0.3385,
+      "step": 25577
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.846681594084104e-07,
+      "loss": 0.3366,
+      "step": 25578
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.842606622385917e-07,
+      "loss": 0.2749,
+      "step": 25579
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83853282074981e-07,
+      "loss": 0.316,
+      "step": 25580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.834460189226943e-07,
+      "loss": 0.3072,
+      "step": 25581
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.830388727868476e-07,
+      "loss": 0.3023,
+      "step": 25582
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.826318436725532e-07,
+      "loss": 0.3249,
+      "step": 25583
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.82224931584925e-07,
+      "loss": 0.3219,
+      "step": 25584
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.818181365290744e-07,
+      "loss": 0.3389,
+      "step": 25585
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.814114585101095e-07,
+      "loss": 0.3002,
+      "step": 25586
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.810048975331374e-07,
+      "loss": 0.314,
+      "step": 25587
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.805984536032672e-07,
+      "loss": 0.2959,
+      "step": 25588
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.801921267256018e-07,
+      "loss": 0.2879,
+      "step": 25589
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.797859169052468e-07,
+      "loss": 0.3077,
+      "step": 25590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.793798241473016e-07,
+      "loss": 0.3046,
+      "step": 25591
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.789738484568675e-07,
+      "loss": 0.293,
+      "step": 25592
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.78567989839043e-07,
+      "loss": 0.3223,
+      "step": 25593
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.781622482989281e-07,
+      "loss": 0.2991,
+      "step": 25594
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.777566238416167e-07,
+      "loss": 0.322,
+      "step": 25595
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.773511164722035e-07,
+      "loss": 0.3293,
+      "step": 25596
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.769457261957823e-07,
+      "loss": 0.2886,
+      "step": 25597
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.765404530174469e-07,
+      "loss": 0.3211,
+      "step": 25598
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.761352969422841e-07,
+      "loss": 0.3127,
+      "step": 25599
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.757302579753854e-07,
+      "loss": 0.3205,
+      "step": 25600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.75325336121837e-07,
+      "loss": 0.3295,
+      "step": 25601
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.749205313867235e-07,
+      "loss": 0.2816,
+      "step": 25602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.745158437751298e-07,
+      "loss": 0.2567,
+      "step": 25603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.741112732921407e-07,
+      "loss": 0.2905,
+      "step": 25604
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.737068199428354e-07,
+      "loss": 0.3032,
+      "step": 25605
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.733024837322944e-07,
+      "loss": 0.322,
+      "step": 25606
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72898264665599e-07,
+      "loss": 0.2864,
+      "step": 25607
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72494162747821e-07,
+      "loss": 0.3365,
+      "step": 25608
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.720901779840394e-07,
+      "loss": 0.3141,
+      "step": 25609
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.716863103793292e-07,
+      "loss": 0.3339,
+      "step": 25610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.712825599387607e-07,
+      "loss": 0.3086,
+      "step": 25611
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.708789266674043e-07,
+      "loss": 0.2952,
+      "step": 25612
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.704754105703337e-07,
+      "loss": 0.2842,
+      "step": 25613
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.700720116526116e-07,
+      "loss": 0.3354,
+      "step": 25614
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.696687299193083e-07,
+      "loss": 0.3219,
+      "step": 25615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.692655653754898e-07,
+      "loss": 0.2854,
+      "step": 25616
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.688625180262165e-07,
+      "loss": 0.3056,
+      "step": 25617
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.684595878765532e-07,
+      "loss": 0.3176,
+      "step": 25618
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.680567749315614e-07,
+      "loss": 0.3147,
+      "step": 25619
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.676540791962971e-07,
+      "loss": 0.2782,
+      "step": 25620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.672515006758217e-07,
+      "loss": 0.31,
+      "step": 25621
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.668490393751891e-07,
+      "loss": 0.2777,
+      "step": 25622
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.664466952994564e-07,
+      "loss": 0.3269,
+      "step": 25623
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.660444684536749e-07,
+      "loss": 0.3128,
+      "step": 25624
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.656423588428995e-07,
+      "loss": 0.3169,
+      "step": 25625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.652403664721763e-07,
+      "loss": 0.2997,
+      "step": 25626
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.648384913465578e-07,
+      "loss": 0.2821,
+      "step": 25627
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.644367334710922e-07,
+      "loss": 0.3107,
+      "step": 25628
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.640350928508221e-07,
+      "loss": 0.3002,
+      "step": 25629
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.636335694907937e-07,
+      "loss": 0.2856,
+      "step": 25630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.632321633960525e-07,
+      "loss": 0.2764,
+      "step": 25631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.628308745716372e-07,
+      "loss": 0.2882,
+      "step": 25632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.624297030225901e-07,
+      "loss": 0.3254,
+      "step": 25633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.620286487539473e-07,
+      "loss": 0.3086,
+      "step": 25634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.616277117707493e-07,
+      "loss": 0.3155,
+      "step": 25635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.612268920780285e-07,
+      "loss": 0.2768,
+      "step": 25636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.608261896808211e-07,
+      "loss": 0.32,
+      "step": 25637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.604256045841618e-07,
+      "loss": 0.2911,
+      "step": 25638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.600251367930776e-07,
+      "loss": 0.3171,
+      "step": 25639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.596247863126016e-07,
+      "loss": 0.2939,
+      "step": 25640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.592245531477626e-07,
+      "loss": 0.3105,
+      "step": 25641
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.588244373035846e-07,
+      "loss": 0.307,
+      "step": 25642
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.584244387850969e-07,
+      "loss": 0.2924,
+      "step": 25643
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.58024557597321e-07,
+      "loss": 0.2892,
+      "step": 25644
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.576247937452785e-07,
+      "loss": 0.3169,
+      "step": 25645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.572251472339919e-07,
+      "loss": 0.3238,
+      "step": 25646
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.568256180684817e-07,
+      "loss": 0.3018,
+      "step": 25647
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.56426206253764e-07,
+      "loss": 0.3649,
+      "step": 25648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.56026911794857e-07,
+      "loss": 0.3293,
+      "step": 25649
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.556277346967754e-07,
+      "loss": 0.3068,
+      "step": 25650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.55228674964532e-07,
+      "loss": 0.3004,
+      "step": 25651
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.548297326031405e-07,
+      "loss": 0.2879,
+      "step": 25652
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.544309076176115e-07,
+      "loss": 0.3081,
+      "step": 25653
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.54032200012954e-07,
+      "loss": 0.3029,
+      "step": 25654
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.536336097941742e-07,
+      "loss": 0.3156,
+      "step": 25655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.532351369662815e-07,
+      "loss": 0.2957,
+      "step": 25656
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.528367815342762e-07,
+      "loss": 0.3517,
+      "step": 25657
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.524385435031655e-07,
+      "loss": 0.3249,
+      "step": 25658
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.52040422877952e-07,
+      "loss": 0.3253,
+      "step": 25659
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.516424196636317e-07,
+      "loss": 0.2953,
+      "step": 25660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.512445338652074e-07,
+      "loss": 0.2851,
+      "step": 25661
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.50846765487676e-07,
+      "loss": 0.2904,
+      "step": 25662
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.504491145360315e-07,
+      "loss": 0.3044,
+      "step": 25663
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.500515810152708e-07,
+      "loss": 0.2947,
+      "step": 25664
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.496541649303867e-07,
+      "loss": 0.3299,
+      "step": 25665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.492568662863685e-07,
+      "loss": 0.3055,
+      "step": 25666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.488596850882068e-07,
+      "loss": 0.3185,
+      "step": 25667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.484626213408929e-07,
+      "loss": 0.3141,
+      "step": 25668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.480656750494119e-07,
+      "loss": 0.3154,
+      "step": 25669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.476688462187486e-07,
+      "loss": 0.2853,
+      "step": 25670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.472721348538891e-07,
+      "loss": 0.3256,
+      "step": 25671
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.46875540959816e-07,
+      "loss": 0.3193,
+      "step": 25672
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.464790645415098e-07,
+      "loss": 0.2759,
+      "step": 25673
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.460827056039509e-07,
+      "loss": 0.3056,
+      "step": 25674
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.456864641521176e-07,
+      "loss": 0.3336,
+      "step": 25675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.452903401909838e-07,
+      "loss": 0.3051,
+      "step": 25676
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.448943337255276e-07,
+      "loss": 0.3308,
+      "step": 25677
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.44498444760725e-07,
+      "loss": 0.2898,
+      "step": 25678
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.441026733015432e-07,
+      "loss": 0.3136,
+      "step": 25679
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.43707019352956e-07,
+      "loss": 0.3151,
+      "step": 25680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.433114829199349e-07,
+      "loss": 0.293,
+      "step": 25681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.429160640074439e-07,
+      "loss": 0.2895,
+      "step": 25682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.42520762620451e-07,
+      "loss": 0.3327,
+      "step": 25683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.421255787639225e-07,
+      "loss": 0.2858,
+      "step": 25684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.417305124428209e-07,
+      "loss": 0.3212,
+      "step": 25685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413355636621077e-07,
+      "loss": 0.3523,
+      "step": 25686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.409407324267448e-07,
+      "loss": 0.288,
+      "step": 25687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.405460187416879e-07,
+      "loss": 0.3029,
+      "step": 25688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.401514226118988e-07,
+      "loss": 0.3084,
+      "step": 25689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.397569440423324e-07,
+      "loss": 0.2958,
+      "step": 25690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.393625830379424e-07,
+      "loss": 0.3144,
+      "step": 25691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.389683396036828e-07,
+      "loss": 0.2967,
+      "step": 25692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.385742137445061e-07,
+      "loss": 0.2941,
+      "step": 25693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.381802054653597e-07,
+      "loss": 0.3116,
+      "step": 25694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.37786314771196e-07,
+      "loss": 0.2941,
+      "step": 25695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.373925416669602e-07,
+      "loss": 0.3007,
+      "step": 25696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.369988861575982e-07,
+      "loss": 0.2617,
+      "step": 25697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.366053482480527e-07,
+      "loss": 0.3314,
+      "step": 25698
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.36211927943271e-07,
+      "loss": 0.2841,
+      "step": 25699
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.358186252481913e-07,
+      "loss": 0.2912,
+      "step": 25700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.354254401677529e-07,
+      "loss": 0.3033,
+      "step": 25701
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.350323727068975e-07,
+      "loss": 0.2949,
+      "step": 25702
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.346394228705577e-07,
+      "loss": 0.3389,
+      "step": 25703
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.342465906636719e-07,
+      "loss": 0.2922,
+      "step": 25704
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.33853876091175e-07,
+      "loss": 0.3138,
+      "step": 25705
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.334612791579975e-07,
+      "loss": 0.3297,
+      "step": 25706
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.330687998690699e-07,
+      "loss": 0.3349,
+      "step": 25707
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.326764382293227e-07,
+      "loss": 0.3446,
+      "step": 25708
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.322841942436853e-07,
+      "loss": 0.3033,
+      "step": 25709
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.318920679170825e-07,
+      "loss": 0.2597,
+      "step": 25710
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.315000592544396e-07,
+      "loss": 0.342,
+      "step": 25711
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.311081682606812e-07,
+      "loss": 0.3395,
+      "step": 25712
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.30716394940728e-07,
+      "loss": 0.3196,
+      "step": 25713
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.303247392995015e-07,
+      "loss": 0.3111,
+      "step": 25714
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.299332013419235e-07,
+      "loss": 0.2899,
+      "step": 25715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.295417810729076e-07,
+      "loss": 0.3108,
+      "step": 25716
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.291504784973712e-07,
+      "loss": 0.2837,
+      "step": 25717
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.287592936202314e-07,
+      "loss": 0.3085,
+      "step": 25718
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.283682264463975e-07,
+      "loss": 0.3089,
+      "step": 25719
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.279772769807845e-07,
+      "loss": 0.329,
+      "step": 25720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.27586445228302e-07,
+      "loss": 0.288,
+      "step": 25721
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.271957311938581e-07,
+      "loss": 0.3418,
+      "step": 25722
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.268051348823612e-07,
+      "loss": 0.2903,
+      "step": 25723
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.264146562987172e-07,
+      "loss": 0.2712,
+      "step": 25724
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.26024295447829e-07,
+      "loss": 0.2604,
+      "step": 25725
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.256340523346016e-07,
+      "loss": 0.2923,
+      "step": 25726
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.252439269639354e-07,
+      "loss": 0.3099,
+      "step": 25727
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.24853919340731e-07,
+      "loss": 0.301,
+      "step": 25728
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.244640294698845e-07,
+      "loss": 0.3171,
+      "step": 25729
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.240742573562974e-07,
+      "loss": 0.3224,
+      "step": 25730
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.236846030048605e-07,
+      "loss": 0.2954,
+      "step": 25731
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.232950664204707e-07,
+      "loss": 0.2999,
+      "step": 25732
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.229056476080208e-07,
+      "loss": 0.3096,
+      "step": 25733
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.225163465724005e-07,
+      "loss": 0.305,
+      "step": 25734
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.221271633185e-07,
+      "loss": 0.2814,
+      "step": 25735
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.217380978512089e-07,
+      "loss": 0.3137,
+      "step": 25736
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.21349150175411e-07,
+      "loss": 0.3381,
+      "step": 25737
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.209603202959936e-07,
+      "loss": 0.3043,
+      "step": 25738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.205716082178415e-07,
+      "loss": 0.2799,
+      "step": 25739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.20183013945832e-07,
+      "loss": 0.3231,
+      "step": 25740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.197945374848502e-07,
+      "loss": 0.31,
+      "step": 25741
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.194061788397743e-07,
+      "loss": 0.3173,
+      "step": 25742
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.190179380154837e-07,
+      "loss": 0.3033,
+      "step": 25743
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.186298150168524e-07,
+      "loss": 0.3056,
+      "step": 25744
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.182418098487553e-07,
+      "loss": 0.3256,
+      "step": 25745
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.178539225160674e-07,
+      "loss": 0.2994,
+      "step": 25746
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.174661530236592e-07,
+      "loss": 0.2755,
+      "step": 25747
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.170785013764024e-07,
+      "loss": 0.3101,
+      "step": 25748
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.166909675791655e-07,
+      "loss": 0.3279,
+      "step": 25749
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.163035516368143e-07,
+      "loss": 0.2735,
+      "step": 25750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.159162535542163e-07,
+      "loss": 0.3127,
+      "step": 25751
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155290733362363e-07,
+      "loss": 0.3105,
+      "step": 25752
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.151420109877348e-07,
+      "loss": 0.3084,
+      "step": 25753
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147550665135771e-07,
+      "loss": 0.3636,
+      "step": 25754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.143682399186213e-07,
+      "loss": 0.2967,
+      "step": 25755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.139815312077247e-07,
+      "loss": 0.2589,
+      "step": 25756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.135949403857466e-07,
+      "loss": 0.3163,
+      "step": 25757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.132084674575422e-07,
+      "loss": 0.3193,
+      "step": 25758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.128221124279665e-07,
+      "loss": 0.2778,
+      "step": 25759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.124358753018688e-07,
+      "loss": 0.3084,
+      "step": 25760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.120497560841033e-07,
+      "loss": 0.3164,
+      "step": 25761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.116637547795168e-07,
+      "loss": 0.3251,
+      "step": 25762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.112778713929601e-07,
+      "loss": 0.3207,
+      "step": 25763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.108921059292805e-07,
+      "loss": 0.3101,
+      "step": 25764
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.105064583933206e-07,
+      "loss": 0.2763,
+      "step": 25765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.101209287899257e-07,
+      "loss": 0.3131,
+      "step": 25766
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.097355171239383e-07,
+      "loss": 0.3251,
+      "step": 25767
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.09350223400198e-07,
+      "loss": 0.2847,
+      "step": 25768
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.089650476235464e-07,
+      "loss": 0.3392,
+      "step": 25769
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.085799897988187e-07,
+      "loss": 0.2883,
+      "step": 25770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.08195049930852e-07,
+      "loss": 0.2892,
+      "step": 25771
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.078102280244813e-07,
+      "loss": 0.3004,
+      "step": 25772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.074255240845417e-07,
+      "loss": 0.2806,
+      "step": 25773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.070409381158615e-07,
+      "loss": 0.3011,
+      "step": 25774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.066564701232736e-07,
+      "loss": 0.2879,
+      "step": 25775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.062721201116073e-07,
+      "loss": 0.2863,
+      "step": 25776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.058878880856889e-07,
+      "loss": 0.2893,
+      "step": 25777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.055037740503433e-07,
+      "loss": 0.2965,
+      "step": 25778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.05119778010399e-07,
+      "loss": 0.3019,
+      "step": 25779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.047358999706765e-07,
+      "loss": 0.29,
+      "step": 25780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.043521399359952e-07,
+      "loss": 0.2958,
+      "step": 25781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.039684979111781e-07,
+      "loss": 0.3171,
+      "step": 25782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.035849739010447e-07,
+      "loss": 0.3224,
+      "step": 25783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.032015679104075e-07,
+      "loss": 0.2843,
+      "step": 25784
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.028182799440851e-07,
+      "loss": 0.2967,
+      "step": 25785
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.024351100068937e-07,
+      "loss": 0.316,
+      "step": 25786
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.020520581036415e-07,
+      "loss": 0.3143,
+      "step": 25787
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.016691242391414e-07,
+      "loss": 0.3142,
+      "step": 25788
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.012863084182064e-07,
+      "loss": 0.3276,
+      "step": 25789
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.009036106456401e-07,
+      "loss": 0.3484,
+      "step": 25790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.00521030926251e-07,
+      "loss": 0.3233,
+      "step": 25791
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.001385692648442e-07,
+      "loss": 0.3079,
+      "step": 25792
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.997562256662226e-07,
+      "loss": 0.2978,
+      "step": 25793
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.993740001351888e-07,
+      "loss": 0.3261,
+      "step": 25794
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.98991892676547e-07,
+      "loss": 0.28,
+      "step": 25795
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.986099032950909e-07,
+      "loss": 0.2866,
+      "step": 25796
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.982280319956202e-07,
+      "loss": 0.2787,
+      "step": 25797
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.978462787829342e-07,
+      "loss": 0.3074,
+      "step": 25798
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.974646436618248e-07,
+      "loss": 0.3169,
+      "step": 25799
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.970831266370869e-07,
+      "loss": 0.3126,
+      "step": 25800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.967017277135112e-07,
+      "loss": 0.2714,
+      "step": 25801
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.963204468958872e-07,
+      "loss": 0.2827,
+      "step": 25802
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.959392841890055e-07,
+      "loss": 0.3302,
+      "step": 25803
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.955582395976545e-07,
+      "loss": 0.317,
+      "step": 25804
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.951773131266169e-07,
+      "loss": 0.3063,
+      "step": 25805
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.947965047806792e-07,
+      "loss": 0.2864,
+      "step": 25806
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.94415814564625e-07,
+      "loss": 0.3366,
+      "step": 25807
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.940352424832341e-07,
+      "loss": 0.301,
+      "step": 25808
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.936547885412869e-07,
+      "loss": 0.3046,
+      "step": 25809
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.932744527435641e-07,
+      "loss": 0.3069,
+      "step": 25810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.928942350948419e-07,
+      "loss": 0.2934,
+      "step": 25811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.92514135599892e-07,
+      "loss": 0.3271,
+      "step": 25812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.92134154263494e-07,
+      "loss": 0.3034,
+      "step": 25813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.917542910904162e-07,
+      "loss": 0.2986,
+      "step": 25814
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.913745460854303e-07,
+      "loss": 0.343,
+      "step": 25815
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.909949192533082e-07,
+      "loss": 0.2741,
+      "step": 25816
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.906154105988182e-07,
+      "loss": 0.3187,
+      "step": 25817
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.902360201267243e-07,
+      "loss": 0.2868,
+      "step": 25818
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.898567478417927e-07,
+      "loss": 0.3004,
+      "step": 25819
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.894775937487884e-07,
+      "loss": 0.3394,
+      "step": 25820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.890985578524733e-07,
+      "loss": 0.3179,
+      "step": 25821
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.887196401576057e-07,
+      "loss": 0.3294,
+      "step": 25822
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.883408406689484e-07,
+      "loss": 0.3474,
+      "step": 25823
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.879621593912554e-07,
+      "loss": 0.3459,
+      "step": 25824
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875835963292842e-07,
+      "loss": 0.2971,
+      "step": 25825
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.872051514877931e-07,
+      "loss": 0.3061,
+      "step": 25826
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.868268248715292e-07,
+      "loss": 0.3305,
+      "step": 25827
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.864486164852489e-07,
+      "loss": 0.3511,
+      "step": 25828
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.860705263337019e-07,
+      "loss": 0.3066,
+      "step": 25829
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.856925544216352e-07,
+      "loss": 0.3081,
+      "step": 25830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.853147007537996e-07,
+      "loss": 0.3293,
+      "step": 25831
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.849369653349357e-07,
+      "loss": 0.3013,
+      "step": 25832
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.845593481697931e-07,
+      "loss": 0.2806,
+      "step": 25833
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.841818492631113e-07,
+      "loss": 0.342,
+      "step": 25834
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.838044686196331e-07,
+      "loss": 0.2828,
+      "step": 25835
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.834272062440971e-07,
+      "loss": 0.3026,
+      "step": 25836
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.830500621412438e-07,
+      "loss": 0.3052,
+      "step": 25837
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.826730363158095e-07,
+      "loss": 0.3004,
+      "step": 25838
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.822961287725282e-07,
+      "loss": 0.3147,
+      "step": 25839
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.819193395161349e-07,
+      "loss": 0.2881,
+      "step": 25840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.815426685513637e-07,
+      "loss": 0.2971,
+      "step": 25841
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.811661158829418e-07,
+      "loss": 0.3186,
+      "step": 25842
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.807896815156033e-07,
+      "loss": 0.3216,
+      "step": 25843
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.804133654540744e-07,
+      "loss": 0.3426,
+      "step": 25844
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.800371677030792e-07,
+      "loss": 0.3153,
+      "step": 25845
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.796610882673448e-07,
+      "loss": 0.3261,
+      "step": 25846
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.792851271515953e-07,
+      "loss": 0.2934,
+      "step": 25847
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.789092843605526e-07,
+      "loss": 0.3152,
+      "step": 25848
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.78533559898935e-07,
+      "loss": 0.327,
+      "step": 25849
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.781579537714666e-07,
+      "loss": 0.3196,
+      "step": 25850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.77782465982859e-07,
+      "loss": 0.3207,
+      "step": 25851
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.774070965378331e-07,
+      "loss": 0.3328,
+      "step": 25852
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.770318454411017e-07,
+      "loss": 0.316,
+      "step": 25853
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.766567126973788e-07,
+      "loss": 0.3059,
+      "step": 25854
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.76281698311374e-07,
+      "loss": 0.2833,
+      "step": 25855
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.759068022877978e-07,
+      "loss": 0.2993,
+      "step": 25856
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.755320246313622e-07,
+      "loss": 0.3199,
+      "step": 25857
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.751573653467712e-07,
+      "loss": 0.3095,
+      "step": 25858
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.74782824438731e-07,
+      "loss": 0.3163,
+      "step": 25859
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.744084019119478e-07,
+      "loss": 0.2714,
+      "step": 25860
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.740340977711223e-07,
+      "loss": 0.2943,
+      "step": 25861
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.736599120209551e-07,
+      "loss": 0.3004,
+      "step": 25862
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.732858446661504e-07,
+      "loss": 0.3125,
+      "step": 25863
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.729118957114021e-07,
+      "loss": 0.2736,
+      "step": 25864
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.725380651614077e-07,
+      "loss": 0.3334,
+      "step": 25865
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.721643530208643e-07,
+      "loss": 0.322,
+      "step": 25866
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.717907592944627e-07,
+      "loss": 0.3201,
+      "step": 25867
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.714172839868982e-07,
+      "loss": 0.3399,
+      "step": 25868
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.710439271028612e-07,
+      "loss": 0.3163,
+      "step": 25869
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.706706886470404e-07,
+      "loss": 0.2932,
+      "step": 25870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.70297568624123e-07,
+      "loss": 0.2987,
+      "step": 25871
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.699245670387976e-07,
+      "loss": 0.3065,
+      "step": 25872
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.69551683895746e-07,
+      "loss": 0.2943,
+      "step": 25873
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.691789191996555e-07,
+      "loss": 0.275,
+      "step": 25874
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.688062729552058e-07,
+      "loss": 0.2712,
+      "step": 25875
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.684337451670762e-07,
+      "loss": 0.3301,
+      "step": 25876
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.680613358399467e-07,
+      "loss": 0.269,
+      "step": 25877
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676890449784966e-07,
+      "loss": 0.3191,
+      "step": 25878
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.67316872587399e-07,
+      "loss": 0.32,
+      "step": 25879
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6694481867133e-07,
+      "loss": 0.3068,
+      "step": 25880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.665728832349638e-07,
+      "loss": 0.2851,
+      "step": 25881
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.662010662829686e-07,
+      "loss": 0.2786,
+      "step": 25882
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.658293678200166e-07,
+      "loss": 0.2774,
+      "step": 25883
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.654577878507773e-07,
+      "loss": 0.2788,
+      "step": 25884
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.650863263799167e-07,
+      "loss": 0.2887,
+      "step": 25885
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.647149834120991e-07,
+      "loss": 0.2925,
+      "step": 25886
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.643437589519907e-07,
+      "loss": 0.3334,
+      "step": 25887
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.639726530042522e-07,
+      "loss": 0.3125,
+      "step": 25888
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.636016655735455e-07,
+      "loss": 0.2993,
+      "step": 25889
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6323079666453e-07,
+      "loss": 0.3106,
+      "step": 25890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.628600462818668e-07,
+      "loss": 0.3285,
+      "step": 25891
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.624894144302073e-07,
+      "loss": 0.2922,
+      "step": 25892
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.621189011142103e-07,
+      "loss": 0.2968,
+      "step": 25893
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.617485063385286e-07,
+      "loss": 0.3229,
+      "step": 25894
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.613782301078153e-07,
+      "loss": 0.3624,
+      "step": 25895
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.610080724267186e-07,
+      "loss": 0.2744,
+      "step": 25896
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.606380332998906e-07,
+      "loss": 0.2915,
+      "step": 25897
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.602681127319754e-07,
+      "loss": 0.3273,
+      "step": 25898
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.598983107276223e-07,
+      "loss": 0.3346,
+      "step": 25899
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.595286272914758e-07,
+      "loss": 0.2954,
+      "step": 25900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.591590624281773e-07,
+      "loss": 0.2951,
+      "step": 25901
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.5878961614237e-07,
+      "loss": 0.3194,
+      "step": 25902
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.584202884386947e-07,
+      "loss": 0.3068,
+      "step": 25903
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.580510793217886e-07,
+      "loss": 0.3359,
+      "step": 25904
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.576819887962914e-07,
+      "loss": 0.3229,
+      "step": 25905
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.57313016866835e-07,
+      "loss": 0.3153,
+      "step": 25906
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.569441635380579e-07,
+      "loss": 0.2874,
+      "step": 25907
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.565754288145886e-07,
+      "loss": 0.2914,
+      "step": 25908
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.562068127010633e-07,
+      "loss": 0.3059,
+      "step": 25909
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.558383152021074e-07,
+      "loss": 0.3332,
+      "step": 25910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.554699363223514e-07,
+      "loss": 0.2919,
+      "step": 25911
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.55101676066424e-07,
+      "loss": 0.2962,
+      "step": 25912
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.54733534438946e-07,
+      "loss": 0.2575,
+      "step": 25913
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.543655114445446e-07,
+      "loss": 0.3084,
+      "step": 25914
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.539976070878427e-07,
+      "loss": 0.2883,
+      "step": 25915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.536298213734582e-07,
+      "loss": 0.3205,
+      "step": 25916
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.532621543060135e-07,
+      "loss": 0.3418,
+      "step": 25917
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.528946058901253e-07,
+      "loss": 0.3104,
+      "step": 25918
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.525271761304085e-07,
+      "loss": 0.3074,
+      "step": 25919
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.521598650314796e-07,
+      "loss": 0.3017,
+      "step": 25920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.517926725979539e-07,
+      "loss": 0.2922,
+      "step": 25921
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.514255988344385e-07,
+      "loss": 0.325,
+      "step": 25922
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.510586437455478e-07,
+      "loss": 0.2768,
+      "step": 25923
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.506918073358913e-07,
+      "loss": 0.2887,
+      "step": 25924
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.503250896100732e-07,
+      "loss": 0.3197,
+      "step": 25925
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.499584905727018e-07,
+      "loss": 0.2946,
+      "step": 25926
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.495920102283814e-07,
+      "loss": 0.2856,
+      "step": 25927
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.492256485817148e-07,
+      "loss": 0.3007,
+      "step": 25928
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.488594056373031e-07,
+      "loss": 0.339,
+      "step": 25929
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.484932813997457e-07,
+      "loss": 0.3198,
+      "step": 25930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.481272758736434e-07,
+      "loss": 0.3297,
+      "step": 25931
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.477613890635902e-07,
+      "loss": 0.3035,
+      "step": 25932
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.473956209741837e-07,
+      "loss": 0.3066,
+      "step": 25933
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.470299716100191e-07,
+      "loss": 0.2768,
+      "step": 25934
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.466644409756861e-07,
+      "loss": 0.3196,
+      "step": 25935
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.462990290757775e-07,
+      "loss": 0.2633,
+      "step": 25936
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.459337359148831e-07,
+      "loss": 0.2803,
+      "step": 25937
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.455685614975914e-07,
+      "loss": 0.29,
+      "step": 25938
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.452035058284855e-07,
+      "loss": 0.2962,
+      "step": 25939
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.44838568912155e-07,
+      "loss": 0.3161,
+      "step": 25940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.444737507531805e-07,
+      "loss": 0.3201,
+      "step": 25941
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.441090513561453e-07,
+      "loss": 0.304,
+      "step": 25942
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.437444707256311e-07,
+      "loss": 0.3143,
+      "step": 25943
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.433800088662145e-07,
+      "loss": 0.3002,
+      "step": 25944
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.430156657824747e-07,
+      "loss": 0.3401,
+      "step": 25945
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.426514414789897e-07,
+      "loss": 0.2965,
+      "step": 25946
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.422873359603309e-07,
+      "loss": 0.2959,
+      "step": 25947
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.419233492310738e-07,
+      "loss": 0.3016,
+      "step": 25948
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.415594812957892e-07,
+      "loss": 0.3091,
+      "step": 25949
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.411957321590456e-07,
+      "loss": 0.3224,
+      "step": 25950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.408321018254137e-07,
+      "loss": 0.2903,
+      "step": 25951
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.404685902994611e-07,
+      "loss": 0.2817,
+      "step": 25952
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.401051975857519e-07,
+      "loss": 0.3193,
+      "step": 25953
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.397419236888513e-07,
+      "loss": 0.3259,
+      "step": 25954
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.393787686133234e-07,
+      "loss": 0.2671,
+      "step": 25955
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.390157323637268e-07,
+      "loss": 0.2943,
+      "step": 25956
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.386528149446224e-07,
+      "loss": 0.3072,
+      "step": 25957
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.382900163605697e-07,
+      "loss": 0.2903,
+      "step": 25958
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.379273366161242e-07,
+      "loss": 0.3159,
+      "step": 25959
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.375647757158408e-07,
+      "loss": 0.2968,
+      "step": 25960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.372023336642729e-07,
+      "loss": 0.2909,
+      "step": 25961
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.368400104659754e-07,
+      "loss": 0.3227,
+      "step": 25962
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.36477806125495e-07,
+      "loss": 0.2839,
+      "step": 25963
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.361157206473844e-07,
+      "loss": 0.3192,
+      "step": 25964
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.357537540361913e-07,
+      "loss": 0.2852,
+      "step": 25965
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.353919062964607e-07,
+      "loss": 0.3078,
+      "step": 25966
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.350301774327371e-07,
+      "loss": 0.2955,
+      "step": 25967
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.346685674495655e-07,
+      "loss": 0.3449,
+      "step": 25968
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.343070763514869e-07,
+      "loss": 0.3392,
+      "step": 25969
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.339457041430408e-07,
+      "loss": 0.3032,
+      "step": 25970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.335844508287679e-07,
+      "loss": 0.3088,
+      "step": 25971
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.332233164132028e-07,
+      "loss": 0.2968,
+      "step": 25972
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.328623009008838e-07,
+      "loss": 0.2993,
+      "step": 25973
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.32501404296345e-07,
+      "loss": 0.2913,
+      "step": 25974
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.321406266041184e-07,
+      "loss": 0.3397,
+      "step": 25975
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.31779967828735e-07,
+      "loss": 0.2883,
+      "step": 25976
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.314194279747276e-07,
+      "loss": 0.3287,
+      "step": 25977
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.310590070466204e-07,
+      "loss": 0.3055,
+      "step": 25978
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.306987050489442e-07,
+      "loss": 0.3237,
+      "step": 25979
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.303385219862223e-07,
+      "loss": 0.315,
+      "step": 25980
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.299784578629774e-07,
+      "loss": 0.2858,
+      "step": 25981
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.296185126837339e-07,
+      "loss": 0.2669,
+      "step": 25982
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.292586864530136e-07,
+      "loss": 0.3035,
+      "step": 25983
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.28898979175333e-07,
+      "loss": 0.3208,
+      "step": 25984
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.285393908552116e-07,
+      "loss": 0.3214,
+      "step": 25985
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.281799214971672e-07,
+      "loss": 0.2893,
+      "step": 25986
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.278205711057127e-07,
+      "loss": 0.3138,
+      "step": 25987
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.274613396853623e-07,
+      "loss": 0.2917,
+      "step": 25988
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.27102227240629e-07,
+      "loss": 0.3174,
+      "step": 25989
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.267432337760226e-07,
+      "loss": 0.3025,
+      "step": 25990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.263843592960494e-07,
+      "loss": 0.3301,
+      "step": 25991
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.260256038052214e-07,
+      "loss": 0.3023,
+      "step": 25992
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.256669673080406e-07,
+      "loss": 0.3792,
+      "step": 25993
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.253084498090144e-07,
+      "loss": 0.336,
+      "step": 25994
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.249500513126449e-07,
+      "loss": 0.3017,
+      "step": 25995
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.245917718234328e-07,
+      "loss": 0.2862,
+      "step": 25996
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.242336113458779e-07,
+      "loss": 0.303,
+      "step": 25997
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.238755698844799e-07,
+      "loss": 0.2946,
+      "step": 25998
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.235176474437376e-07,
+      "loss": 0.2985,
+      "step": 25999
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.231598440281438e-07,
+      "loss": 0.2932,
+      "step": 26000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.228021596421917e-07,
+      "loss": 0.3122,
+      "step": 26001
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.224445942903766e-07,
+      "loss": 0.2964,
+      "step": 26002
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.220871479771872e-07,
+      "loss": 0.2957,
+      "step": 26003
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.21729820707113e-07,
+      "loss": 0.3136,
+      "step": 26004
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.213726124846452e-07,
+      "loss": 0.3018,
+      "step": 26005
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210155233142666e-07,
+      "loss": 0.3382,
+      "step": 26006
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.206585532004649e-07,
+      "loss": 0.3046,
+      "step": 26007
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.20301702147723e-07,
+      "loss": 0.2745,
+      "step": 26008
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.199449701605219e-07,
+      "loss": 0.3026,
+      "step": 26009
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.195883572433435e-07,
+      "loss": 0.3141,
+      "step": 26010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.192318634006654e-07,
+      "loss": 0.3109,
+      "step": 26011
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.188754886369673e-07,
+      "loss": 0.3092,
+      "step": 26012
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.185192329567235e-07,
+      "loss": 0.3172,
+      "step": 26013
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.181630963644102e-07,
+      "loss": 0.2449,
+      "step": 26014
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.178070788644973e-07,
+      "loss": 0.3096,
+      "step": 26015
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.174511804614591e-07,
+      "loss": 0.3083,
+      "step": 26016
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.170954011597662e-07,
+      "loss": 0.3371,
+      "step": 26017
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.167397409638853e-07,
+      "loss": 0.3002,
+      "step": 26018
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.163841998782837e-07,
+      "loss": 0.3429,
+      "step": 26019
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.160287779074302e-07,
+      "loss": 0.2869,
+      "step": 26020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.156734750557835e-07,
+      "loss": 0.2806,
+      "step": 26021
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.153182913278121e-07,
+      "loss": 0.3023,
+      "step": 26022
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.149632267279736e-07,
+      "loss": 0.3006,
+      "step": 26023
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146082812607256e-07,
+      "loss": 0.3138,
+      "step": 26024
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.1425345493053e-07,
+      "loss": 0.3192,
+      "step": 26025
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.138987477418434e-07,
+      "loss": 0.3016,
+      "step": 26026
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.135441596991187e-07,
+      "loss": 0.2952,
+      "step": 26027
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.131896908068101e-07,
+      "loss": 0.2853,
+      "step": 26028
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.128353410693721e-07,
+      "loss": 0.3349,
+      "step": 26029
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.12481110491253e-07,
+      "loss": 0.3031,
+      "step": 26030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.121269990769007e-07,
+      "loss": 0.3133,
+      "step": 26031
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.117730068307669e-07,
+      "loss": 0.2954,
+      "step": 26032
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.11419133757296e-07,
+      "loss": 0.3074,
+      "step": 26033
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.110653798609311e-07,
+      "loss": 0.3007,
+      "step": 26034
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.107117451461152e-07,
+      "loss": 0.3233,
+      "step": 26035
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.103582296172938e-07,
+      "loss": 0.3072,
+      "step": 26036
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.100048332789021e-07,
+      "loss": 0.311,
+      "step": 26037
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.096515561353821e-07,
+      "loss": 0.2948,
+      "step": 26038
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.092983981911703e-07,
+      "loss": 0.3355,
+      "step": 26039
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.089453594507021e-07,
+      "loss": 0.3053,
+      "step": 26040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.085924399184106e-07,
+      "loss": 0.2969,
+      "step": 26041
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.082396395987321e-07,
+      "loss": 0.2771,
+      "step": 26042
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078869584960943e-07,
+      "loss": 0.3304,
+      "step": 26043
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.075343966149271e-07,
+      "loss": 0.327,
+      "step": 26044
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.071819539596612e-07,
+      "loss": 0.3191,
+      "step": 26045
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.068296305347198e-07,
+      "loss": 0.3438,
+      "step": 26046
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.064774263445294e-07,
+      "loss": 0.3005,
+      "step": 26047
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.061253413935163e-07,
+      "loss": 0.3063,
+      "step": 26048
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.057733756860994e-07,
+      "loss": 0.3443,
+      "step": 26049
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.054215292267007e-07,
+      "loss": 0.3034,
+      "step": 26050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.050698020197398e-07,
+      "loss": 0.2944,
+      "step": 26051
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.047181940696333e-07,
+      "loss": 0.3042,
+      "step": 26052
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043667053807988e-07,
+      "loss": 0.2684,
+      "step": 26053
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.040153359576516e-07,
+      "loss": 0.3404,
+      "step": 26054
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.036640858046005e-07,
+      "loss": 0.3322,
+      "step": 26055
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.033129549260618e-07,
+      "loss": 0.3128,
+      "step": 26056
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.029619433264454e-07,
+      "loss": 0.2843,
+      "step": 26057
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.026110510101567e-07,
+      "loss": 0.2938,
+      "step": 26058
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.022602779816055e-07,
+      "loss": 0.3314,
+      "step": 26059
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.019096242451993e-07,
+      "loss": 0.326,
+      "step": 26060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.01559089805338e-07,
+      "loss": 0.2929,
+      "step": 26061
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.012086746664269e-07,
+      "loss": 0.2779,
+      "step": 26062
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.008583788328681e-07,
+      "loss": 0.3006,
+      "step": 26063
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.00508202309059e-07,
+      "loss": 0.3303,
+      "step": 26064
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.001581450993987e-07,
+      "loss": 0.34,
+      "step": 26065
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.998082072082855e-07,
+      "loss": 0.3211,
+      "step": 26066
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.994583886401117e-07,
+      "loss": 0.3003,
+      "step": 26067
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.991086893992725e-07,
+      "loss": 0.3088,
+      "step": 26068
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.987591094901623e-07,
+      "loss": 0.2808,
+      "step": 26069
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.984096489171675e-07,
+      "loss": 0.2638,
+      "step": 26070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.980603076846802e-07,
+      "loss": 0.2575,
+      "step": 26071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.977110857970868e-07,
+      "loss": 0.2985,
+      "step": 26072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.973619832587773e-07,
+      "loss": 0.304,
+      "step": 26073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.970130000741324e-07,
+      "loss": 0.3203,
+      "step": 26074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.966641362475355e-07,
+      "loss": 0.3045,
+      "step": 26075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.963153917833696e-07,
+      "loss": 0.3226,
+      "step": 26076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.959667666860147e-07,
+      "loss": 0.3136,
+      "step": 26077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.956182609598492e-07,
+      "loss": 0.3143,
+      "step": 26078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.95269874609251e-07,
+      "loss": 0.3165,
+      "step": 26079
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.949216076385954e-07,
+      "loss": 0.3043,
+      "step": 26080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.945734600522555e-07,
+      "loss": 0.2741,
+      "step": 26081
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.942254318546081e-07,
+      "loss": 0.3054,
+      "step": 26082
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.938775230500192e-07,
+      "loss": 0.3017,
+      "step": 26083
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.935297336428635e-07,
+      "loss": 0.3031,
+      "step": 26084
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.931820636375051e-07,
+      "loss": 0.2956,
+      "step": 26085
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.928345130383128e-07,
+      "loss": 0.3175,
+      "step": 26086
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.924870818496496e-07,
+      "loss": 0.303,
+      "step": 26087
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921397700758845e-07,
+      "loss": 0.3186,
+      "step": 26088
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.917925777213739e-07,
+      "loss": 0.3001,
+      "step": 26089
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.91445504790481e-07,
+      "loss": 0.2904,
+      "step": 26090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.910985512875666e-07,
+      "loss": 0.3108,
+      "step": 26091
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.907517172169862e-07,
+      "loss": 0.294,
+      "step": 26092
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904050025830965e-07,
+      "loss": 0.3314,
+      "step": 26093
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.900584073902526e-07,
+      "loss": 0.2849,
+      "step": 26094
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.89711931642809e-07,
+      "loss": 0.2993,
+      "step": 26095
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.893655753451154e-07,
+      "loss": 0.3089,
+      "step": 26096
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.89019338501523e-07,
+      "loss": 0.3232,
+      "step": 26097
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.886732211163792e-07,
+      "loss": 0.3086,
+      "step": 26098
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.883272231940328e-07,
+      "loss": 0.3535,
+      "step": 26099
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.879813447388304e-07,
+      "loss": 0.266,
+      "step": 26100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.87635585755114e-07,
+      "loss": 0.3332,
+      "step": 26101
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.872899462472269e-07,
+      "loss": 0.3003,
+      "step": 26102
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.869444262195122e-07,
+      "loss": 0.3123,
+      "step": 26103
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.865990256763076e-07,
+      "loss": 0.3021,
+      "step": 26104
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.862537446219528e-07,
+      "loss": 0.2984,
+      "step": 26105
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.859085830607823e-07,
+      "loss": 0.2748,
+      "step": 26106
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.855635409971349e-07,
+      "loss": 0.3071,
+      "step": 26107
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.852186184353402e-07,
+      "loss": 0.3294,
+      "step": 26108
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.848738153797328e-07,
+      "loss": 0.309,
+      "step": 26109
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.845291318346445e-07,
+      "loss": 0.3287,
+      "step": 26110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.841845678044022e-07,
+      "loss": 0.316,
+      "step": 26111
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.838401232933343e-07,
+      "loss": 0.3088,
+      "step": 26112
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.834957983057698e-07,
+      "loss": 0.2779,
+      "step": 26113
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.831515928460284e-07,
+      "loss": 0.2921,
+      "step": 26114
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.828075069184379e-07,
+      "loss": 0.2917,
+      "step": 26115
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.82463540527317e-07,
+      "loss": 0.2951,
+      "step": 26116
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.821196936769879e-07,
+      "loss": 0.3283,
+      "step": 26117
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.817759663717669e-07,
+      "loss": 0.3282,
+      "step": 26118
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.814323586159753e-07,
+      "loss": 0.3139,
+      "step": 26119
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.810888704139238e-07,
+      "loss": 0.3032,
+      "step": 26120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.807455017699303e-07,
+      "loss": 0.3073,
+      "step": 26121
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804022526883067e-07,
+      "loss": 0.3245,
+      "step": 26122
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.800591231733631e-07,
+      "loss": 0.3045,
+      "step": 26123
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.797161132294103e-07,
+      "loss": 0.3267,
+      "step": 26124
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.793732228607573e-07,
+      "loss": 0.3165,
+      "step": 26125
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.790304520717093e-07,
+      "loss": 0.329,
+      "step": 26126
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.78687800866573e-07,
+      "loss": 0.3107,
+      "step": 26127
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.783452692496504e-07,
+      "loss": 0.3396,
+      "step": 26128
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.780028572252437e-07,
+      "loss": 0.3057,
+      "step": 26129
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.776605647976551e-07,
+      "loss": 0.3239,
+      "step": 26130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.773183919711833e-07,
+      "loss": 0.2943,
+      "step": 26131
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769763387501247e-07,
+      "loss": 0.3143,
+      "step": 26132
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7663440513877723e-07,
+      "loss": 0.3031,
+      "step": 26133
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.762925911414362e-07,
+      "loss": 0.3436,
+      "step": 26134
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.759508967623916e-07,
+      "loss": 0.3092,
+      "step": 26135
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.756093220059377e-07,
+      "loss": 0.3012,
+      "step": 26136
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.752678668763655e-07,
+      "loss": 0.3507,
+      "step": 26137
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.749265313779627e-07,
+      "loss": 0.331,
+      "step": 26138
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7458531551501484e-07,
+      "loss": 0.3199,
+      "step": 26139
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.742442192918095e-07,
+      "loss": 0.2974,
+      "step": 26140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7390324271262887e-07,
+      "loss": 0.295,
+      "step": 26141
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.735623857817573e-07,
+      "loss": 0.2915,
+      "step": 26142
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7322164850347685e-07,
+      "loss": 0.2992,
+      "step": 26143
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.728810308820653e-07,
+      "loss": 0.3243,
+      "step": 26144
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7254053292180133e-07,
+      "loss": 0.3396,
+      "step": 26145
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7220015462696165e-07,
+      "loss": 0.2744,
+      "step": 26146
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.718598960018228e-07,
+      "loss": 0.2681,
+      "step": 26147
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.71519757050658e-07,
+      "loss": 0.3059,
+      "step": 26148
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.711797377777372e-07,
+      "loss": 0.3212,
+      "step": 26149
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.708398381873336e-07,
+      "loss": 0.3129,
+      "step": 26150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7050005828371383e-07,
+      "loss": 0.3183,
+      "step": 26151
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.701603980711478e-07,
+      "loss": 0.3369,
+      "step": 26152
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6982085755390094e-07,
+      "loss": 0.2829,
+      "step": 26153
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.694814367362377e-07,
+      "loss": 0.285,
+      "step": 26154
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.691421356224213e-07,
+      "loss": 0.2934,
+      "step": 26155
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6880295421671384e-07,
+      "loss": 0.3175,
+      "step": 26156
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6846389252337423e-07,
+      "loss": 0.3019,
+      "step": 26157
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.681249505466634e-07,
+      "loss": 0.3068,
+      "step": 26158
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6778612829083693e-07,
+      "loss": 0.3082,
+      "step": 26159
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6744742576015023e-07,
+      "loss": 0.2998,
+      "step": 26160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6710884295885664e-07,
+      "loss": 0.2688,
+      "step": 26161
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6677037989121153e-07,
+      "loss": 0.2913,
+      "step": 26162
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.664320365614627e-07,
+      "loss": 0.3245,
+      "step": 26163
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.660938129738612e-07,
+      "loss": 0.2936,
+      "step": 26164
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.657557091326581e-07,
+      "loss": 0.332,
+      "step": 26165
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6541772504209436e-07,
+      "loss": 0.3054,
+      "step": 26166
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6507986070641886e-07,
+      "loss": 0.2876,
+      "step": 26167
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6474211612987484e-07,
+      "loss": 0.2815,
+      "step": 26168
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.644044913167045e-07,
+      "loss": 0.2858,
+      "step": 26169
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6406698627114665e-07,
+      "loss": 0.3387,
+      "step": 26170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6372960099744346e-07,
+      "loss": 0.2999,
+      "step": 26171
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.633923354998282e-07,
+      "loss": 0.3222,
+      "step": 26172
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6305518978253974e-07,
+      "loss": 0.3362,
+      "step": 26173
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.627181638498135e-07,
+      "loss": 0.279,
+      "step": 26174
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.623812577058795e-07,
+      "loss": 0.3138,
+      "step": 26175
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.62044471354971e-07,
+      "loss": 0.2958,
+      "step": 26176
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6170780480132017e-07,
+      "loss": 0.3177,
+      "step": 26177
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.613712580491514e-07,
+      "loss": 0.3052,
+      "step": 26178
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6103483110269575e-07,
+      "loss": 0.2929,
+      "step": 26179
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.606985239661754e-07,
+      "loss": 0.3182,
+      "step": 26180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.60362336643817e-07,
+      "loss": 0.3318,
+      "step": 26181
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.600262691398416e-07,
+      "loss": 0.3538,
+      "step": 26182
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5969032145846914e-07,
+      "loss": 0.2787,
+      "step": 26183
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5935449360392294e-07,
+      "loss": 0.3395,
+      "step": 26184
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5901878558041623e-07,
+      "loss": 0.3161,
+      "step": 26185
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5868319739216903e-07,
+      "loss": 0.3498,
+      "step": 26186
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5834772904339577e-07,
+      "loss": 0.2982,
+      "step": 26187
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5801238053830856e-07,
+      "loss": 0.2967,
+      "step": 26188
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5767715188112184e-07,
+      "loss": 0.3208,
+      "step": 26189
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5734204307604227e-07,
+      "loss": 0.3064,
+      "step": 26190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5700705412728305e-07,
+      "loss": 0.2958,
+      "step": 26191
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5667218503904876e-07,
+      "loss": 0.2997,
+      "step": 26192
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.56337435815547e-07,
+      "loss": 0.2692,
+      "step": 26193
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.560028064609801e-07,
+      "loss": 0.2906,
+      "step": 26194
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.556682969795534e-07,
+      "loss": 0.2967,
+      "step": 26195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5533390737546813e-07,
+      "loss": 0.319,
+      "step": 26196
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54999637652922e-07,
+      "loss": 0.3196,
+      "step": 26197
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.54665487816115e-07,
+      "loss": 0.2914,
+      "step": 26198
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.543314578692459e-07,
+      "loss": 0.3109,
+      "step": 26199
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5399754781650594e-07,
+      "loss": 0.3006,
+      "step": 26200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.536637576620939e-07,
+      "loss": 0.3113,
+      "step": 26201
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.533300874101998e-07,
+      "loss": 0.2976,
+      "step": 26202
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5299653706501246e-07,
+      "loss": 0.2936,
+      "step": 26203
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.52663106630723e-07,
+      "loss": 0.267,
+      "step": 26204
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.523297961115214e-07,
+      "loss": 0.3543,
+      "step": 26205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5199660551159095e-07,
+      "loss": 0.3246,
+      "step": 26206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.516635348351184e-07,
+      "loss": 0.3214,
+      "step": 26207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5133058408628805e-07,
+      "loss": 0.2973,
+      "step": 26208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5099775326927776e-07,
+      "loss": 0.2642,
+      "step": 26209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5066504238827303e-07,
+      "loss": 0.284,
+      "step": 26210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.503324514474483e-07,
+      "loss": 0.2843,
+      "step": 26211
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.499999804509836e-07,
+      "loss": 0.274,
+      "step": 26212
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4966762940305333e-07,
+      "loss": 0.3088,
+      "step": 26213
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.493353983078341e-07,
+      "loss": 0.2946,
+      "step": 26214
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4900328716949603e-07,
+      "loss": 0.3133,
+      "step": 26215
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4867129599221015e-07,
+      "loss": 0.3317,
+      "step": 26216
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.483394247801487e-07,
+      "loss": 0.3228,
+      "step": 26217
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4800767353748055e-07,
+      "loss": 0.3007,
+      "step": 26218
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4767604226837015e-07,
+      "loss": 0.3087,
+      "step": 26219
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4734453097698416e-07,
+      "loss": 0.2702,
+      "step": 26220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.470131396674848e-07,
+      "loss": 0.3025,
+      "step": 26221
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.466818683440366e-07,
+      "loss": 0.2654,
+      "step": 26222
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4635071701079726e-07,
+      "loss": 0.3313,
+      "step": 26223
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.460196856719301e-07,
+      "loss": 0.3105,
+      "step": 26224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4568877433158964e-07,
+      "loss": 0.2832,
+      "step": 26225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.453579829939325e-07,
+      "loss": 0.3413,
+      "step": 26226
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.450273116631154e-07,
+      "loss": 0.331,
+      "step": 26227
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4469676034329055e-07,
+      "loss": 0.3108,
+      "step": 26228
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.44366329038608e-07,
+      "loss": 0.3286,
+      "step": 26229
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4403601775322214e-07,
+      "loss": 0.2817,
+      "step": 26230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4370582649127635e-07,
+      "loss": 0.3232,
+      "step": 26231
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.433757552569229e-07,
+      "loss": 0.3028,
+      "step": 26232
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.430458040543051e-07,
+      "loss": 0.2744,
+      "step": 26233
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.427159728875652e-07,
+      "loss": 0.3111,
+      "step": 26234
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4238626176084874e-07,
+      "loss": 0.3161,
+      "step": 26235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4205667067829696e-07,
+      "loss": 0.3017,
+      "step": 26236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.417271996440486e-07,
+      "loss": 0.3354,
+      "step": 26237
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.413978486622406e-07,
+      "loss": 0.3204,
+      "step": 26238
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.410686177370127e-07,
+      "loss": 0.3312,
+      "step": 26239
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.407395068724973e-07,
+      "loss": 0.3167,
+      "step": 26240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.404105160728278e-07,
+      "loss": 0.3224,
+      "step": 26241
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.400816453421408e-07,
+      "loss": 0.306,
+      "step": 26242
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3975289468456194e-07,
+      "loss": 0.2936,
+      "step": 26243
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3942426410422233e-07,
+      "loss": 0.3082,
+      "step": 26244
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.390957536052498e-07,
+      "loss": 0.2758,
+      "step": 26245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.387673631917688e-07,
+      "loss": 0.3124,
+      "step": 26246
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.384390928679061e-07,
+      "loss": 0.3072,
+      "step": 26247
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3811094263778386e-07,
+      "loss": 0.2997,
+      "step": 26248
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.377829125055233e-07,
+      "loss": 0.3207,
+      "step": 26249
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3745500247524555e-07,
+      "loss": 0.2913,
+      "step": 26250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3712721255106835e-07,
+      "loss": 0.3031,
+      "step": 26251
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.367995427371086e-07,
+      "loss": 0.3082,
+      "step": 26252
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3647199303748276e-07,
+      "loss": 0.277,
+      "step": 26253
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.361445634563033e-07,
+      "loss": 0.3075,
+      "step": 26254
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3581725399768573e-07,
+      "loss": 0.3193,
+      "step": 26255
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3549006466573784e-07,
+      "loss": 0.2752,
+      "step": 26256
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.351629954645697e-07,
+      "loss": 0.3116,
+      "step": 26257
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3483604639829146e-07,
+      "loss": 0.3141,
+      "step": 26258
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.345092174710064e-07,
+      "loss": 0.3292,
+      "step": 26259
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.341825086868223e-07,
+      "loss": 0.3134,
+      "step": 26260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3385592004984155e-07,
+      "loss": 0.2423,
+      "step": 26261
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3352945156416636e-07,
+      "loss": 0.3104,
+      "step": 26262
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.332031032338979e-07,
+      "loss": 0.2937,
+      "step": 26263
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.32876875063134e-07,
+      "loss": 0.3015,
+      "step": 26264
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3255076705597143e-07,
+      "loss": 0.3127,
+      "step": 26265
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.322247792165069e-07,
+      "loss": 0.2931,
+      "step": 26266
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.318989115488359e-07,
+      "loss": 0.3013,
+      "step": 26267
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3157316405704977e-07,
+      "loss": 0.2764,
+      "step": 26268
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3124753674523955e-07,
+      "loss": 0.3155,
+      "step": 26269
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.309220296174987e-07,
+      "loss": 0.3204,
+      "step": 26270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.305966426779118e-07,
+      "loss": 0.3257,
+      "step": 26271
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3027137593056655e-07,
+      "loss": 0.3085,
+      "step": 26272
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.299462293795498e-07,
+      "loss": 0.3078,
+      "step": 26273
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2962120302894484e-07,
+      "loss": 0.3055,
+      "step": 26274
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.292962968828318e-07,
+      "loss": 0.3335,
+      "step": 26275
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.289715109452952e-07,
+      "loss": 0.2903,
+      "step": 26276
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.286468452204107e-07,
+      "loss": 0.3315,
+      "step": 26277
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2832229971225824e-07,
+      "loss": 0.2843,
+      "step": 26278
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.279978744249147e-07,
+      "loss": 0.3148,
+      "step": 26279
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2767356936245233e-07,
+      "loss": 0.307,
+      "step": 26280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.273493845289467e-07,
+      "loss": 0.3029,
+      "step": 26281
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.270253199284702e-07,
+      "loss": 0.3511,
+      "step": 26282
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.267013755650906e-07,
+      "loss": 0.2891,
+      "step": 26283
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2637755144287805e-07,
+      "loss": 0.3169,
+      "step": 26284
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.260538475658993e-07,
+      "loss": 0.3091,
+      "step": 26285
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.257302639382221e-07,
+      "loss": 0.3077,
+      "step": 26286
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.254068005639067e-07,
+      "loss": 0.2909,
+      "step": 26287
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2508345744702083e-07,
+      "loss": 0.2893,
+      "step": 26288
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2476023459162017e-07,
+      "loss": 0.3318,
+      "step": 26289
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2443713200176816e-07,
+      "loss": 0.303,
+      "step": 26290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.241141496815215e-07,
+      "loss": 0.3049,
+      "step": 26291
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2379128763493925e-07,
+      "loss": 0.298,
+      "step": 26292
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2346854586607365e-07,
+      "loss": 0.3061,
+      "step": 26293
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.231459243789804e-07,
+      "loss": 0.3114,
+      "step": 26294
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.228234231777095e-07,
+      "loss": 0.3181,
+      "step": 26295
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.225010422663134e-07,
+      "loss": 0.2982,
+      "step": 26296
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2217878164883985e-07,
+      "loss": 0.3169,
+      "step": 26297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.21856641329339e-07,
+      "loss": 0.3316,
+      "step": 26298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.215346213118532e-07,
+      "loss": 0.31,
+      "step": 26299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.212127216004291e-07,
+      "loss": 0.3086,
+      "step": 26300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2089094219911143e-07,
+      "loss": 0.2881,
+      "step": 26301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.205692831119379e-07,
+      "loss": 0.3012,
+      "step": 26302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2024774434295204e-07,
+      "loss": 0.2927,
+      "step": 26303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1992632589619054e-07,
+      "loss": 0.2917,
+      "step": 26304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1960502777569133e-07,
+      "loss": 0.3281,
+      "step": 26305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.192838499854901e-07,
+      "loss": 0.2889,
+      "step": 26306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.189627925296202e-07,
+      "loss": 0.3089,
+      "step": 26307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.18641855412113e-07,
+      "loss": 0.3306,
+      "step": 26308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.183210386370007e-07,
+      "loss": 0.3122,
+      "step": 26309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1800034220831344e-07,
+      "loss": 0.2987,
+      "step": 26310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1767976613007797e-07,
+      "loss": 0.2989,
+      "step": 26311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.173593104063212e-07,
+      "loss": 0.3056,
+      "step": 26312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1703897504106974e-07,
+      "loss": 0.305,
+      "step": 26313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.167187600383438e-07,
+      "loss": 0.3258,
+      "step": 26314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1639866540216677e-07,
+      "loss": 0.3459,
+      "step": 26315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1607869113656105e-07,
+      "loss": 0.3162,
+      "step": 26316
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1575883724554344e-07,
+      "loss": 0.2398,
+      "step": 26317
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.154391037331307e-07,
+      "loss": 0.3144,
+      "step": 26318
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1511949060334065e-07,
+      "loss": 0.2919,
+      "step": 26319
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1479999786018464e-07,
+      "loss": 0.3142,
+      "step": 26320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1448062550767834e-07,
+      "loss": 0.3115,
+      "step": 26321
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1416137354983397e-07,
+      "loss": 0.3166,
+      "step": 26322
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1384224199065737e-07,
+      "loss": 0.3194,
+      "step": 26323
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1352323083415966e-07,
+      "loss": 0.3341,
+      "step": 26324
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1320434008434774e-07,
+      "loss": 0.2718,
+      "step": 26325
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.12885569745225e-07,
+      "loss": 0.3266,
+      "step": 26326
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.125669198207982e-07,
+      "loss": 0.3229,
+      "step": 26327
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.122483903150665e-07,
+      "loss": 0.3057,
+      "step": 26328
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.119299812320332e-07,
+      "loss": 0.2965,
+      "step": 26329
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.116116925756952e-07,
+      "loss": 0.3025,
+      "step": 26330
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1129352435005044e-07,
+      "loss": 0.3387,
+      "step": 26331
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1097547655909785e-07,
+      "loss": 0.2711,
+      "step": 26332
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.106575492068299e-07,
+      "loss": 0.3363,
+      "step": 26333
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.103397422972388e-07,
+      "loss": 0.3285,
+      "step": 26334
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.100220558343204e-07,
+      "loss": 0.3042,
+      "step": 26335
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0970448982205924e-07,
+      "loss": 0.3105,
+      "step": 26336
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0938704426444873e-07,
+      "loss": 0.3131,
+      "step": 26337
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0906971916547465e-07,
+      "loss": 0.2812,
+      "step": 26338
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.087525145291205e-07,
+      "loss": 0.2658,
+      "step": 26339
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0843543035937185e-07,
+      "loss": 0.2846,
+      "step": 26340
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.081184666602123e-07,
+      "loss": 0.3268,
+      "step": 26341
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.078016234356208e-07,
+      "loss": 0.2864,
+      "step": 26342
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.074849006895787e-07,
+      "loss": 0.3047,
+      "step": 26343
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0716829842606385e-07,
+      "loss": 0.2929,
+      "step": 26344
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0685181664905206e-07,
+      "loss": 0.2657,
+      "step": 26345
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0653545536251895e-07,
+      "loss": 0.3373,
+      "step": 26346
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.06219214570438e-07,
+      "loss": 0.26,
+      "step": 26347
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.059030942767817e-07,
+      "loss": 0.2748,
+      "step": 26348
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.055870944855178e-07,
+      "loss": 0.3073,
+      "step": 26349
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0527121520061995e-07,
+      "loss": 0.3028,
+      "step": 26350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.049554564260505e-07,
+      "loss": 0.307,
+      "step": 26351
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0463981816577846e-07,
+      "loss": 0.2952,
+      "step": 26352
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0432430042376957e-07,
+      "loss": 0.3022,
+      "step": 26353
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.040089032039829e-07,
+      "loss": 0.328,
+      "step": 26354
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0369362651038193e-07,
+      "loss": 0.3424,
+      "step": 26355
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0337847034692787e-07,
+      "loss": 0.3447,
+      "step": 26356
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.030634347175755e-07,
+      "loss": 0.3302,
+      "step": 26357
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.027485196262859e-07,
+      "loss": 0.3452,
+      "step": 26358
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.024337250770127e-07,
+      "loss": 0.2993,
+      "step": 26359
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.021190510737083e-07,
+      "loss": 0.3055,
+      "step": 26360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.01804497620325e-07,
+      "loss": 0.2888,
+      "step": 26361
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.014900647208175e-07,
+      "loss": 0.3276,
+      "step": 26362
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.011757523791304e-07,
+      "loss": 0.2843,
+      "step": 26363
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0086156059921276e-07,
+      "loss": 0.3324,
+      "step": 26364
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0054748938501255e-07,
+      "loss": 0.3049,
+      "step": 26365
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0023353874047435e-07,
+      "loss": 0.3308,
+      "step": 26366
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9991970866953947e-07,
+      "loss": 0.28,
+      "step": 26367
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.996059991761514e-07,
+      "loss": 0.2944,
+      "step": 26368
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.992924102642504e-07,
+      "loss": 0.3091,
+      "step": 26369
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.989789419377732e-07,
+      "loss": 0.3134,
+      "step": 26370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.986655942006579e-07,
+      "loss": 0.3163,
+      "step": 26371
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9835236705684123e-07,
+      "loss": 0.2743,
+      "step": 26372
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9803926051025566e-07,
+      "loss": 0.304,
+      "step": 26373
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.977262745648358e-07,
+      "loss": 0.3112,
+      "step": 26374
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.974134092245119e-07,
+      "loss": 0.3119,
+      "step": 26375
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9710066449321183e-07,
+      "loss": 0.3036,
+      "step": 26376
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9678804037486584e-07,
+      "loss": 0.3174,
+      "step": 26377
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.964755368734008e-07,
+      "loss": 0.322,
+      "step": 26378
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.961631539927413e-07,
+      "loss": 0.3291,
+      "step": 26379
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.958508917368087e-07,
+      "loss": 0.3135,
+      "step": 26380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.955387501095287e-07,
+      "loss": 0.2934,
+      "step": 26381
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.952267291148182e-07,
+      "loss": 0.3421,
+      "step": 26382
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.949148287565985e-07,
+      "loss": 0.2762,
+      "step": 26383
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9460304903878754e-07,
+      "loss": 0.2844,
+      "step": 26384
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9429138996529783e-07,
+      "loss": 0.316,
+      "step": 26385
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9397985154004727e-07,
+      "loss": 0.2872,
+      "step": 26386
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.936684337669494e-07,
+      "loss": 0.3162,
+      "step": 26387
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9335713664991224e-07,
+      "loss": 0.2542,
+      "step": 26388
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.930459601928482e-07,
+      "loss": 0.3445,
+      "step": 26389
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9273490439966424e-07,
+      "loss": 0.295,
+      "step": 26390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.924239692742693e-07,
+      "loss": 0.3182,
+      "step": 26391
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.921131548205659e-07,
+      "loss": 0.3207,
+      "step": 26392
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.918024610424609e-07,
+      "loss": 0.3099,
+      "step": 26393
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.914918879438545e-07,
+      "loss": 0.3419,
+      "step": 26394
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.91181435528647e-07,
+      "loss": 0.2936,
+      "step": 26395
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.908711038007407e-07,
+      "loss": 0.2919,
+      "step": 26396
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.905608927640292e-07,
+      "loss": 0.2792,
+      "step": 26397
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.902508024224128e-07,
+      "loss": 0.2791,
+      "step": 26398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.899408327797849e-07,
+      "loss": 0.322,
+      "step": 26399
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.896309838400369e-07,
+      "loss": 0.3043,
+      "step": 26400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8932125560706247e-07,
+      "loss": 0.2953,
+      "step": 26401
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.890116480847517e-07,
+      "loss": 0.3302,
+      "step": 26402
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.887021612769937e-07,
+      "loss": 0.3226,
+      "step": 26403
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8839279518767315e-07,
+      "loss": 0.3126,
+      "step": 26404
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8808354982067696e-07,
+      "loss": 0.3076,
+      "step": 26405
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8777442517989206e-07,
+      "loss": 0.2952,
+      "step": 26406
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8746542126919753e-07,
+      "loss": 0.311,
+      "step": 26407
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.871565380924747e-07,
+      "loss": 0.3117,
+      "step": 26408
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8684777565360596e-07,
+      "loss": 0.3002,
+      "step": 26409
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8653913395646616e-07,
+      "loss": 0.358,
+      "step": 26410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8623061300493427e-07,
+      "loss": 0.3224,
+      "step": 26411
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.85922212802885e-07,
+      "loss": 0.271,
+      "step": 26412
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.856139333541886e-07,
+      "loss": 0.3057,
+      "step": 26413
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.853057746627209e-07,
+      "loss": 0.264,
+      "step": 26414
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8499773673235206e-07,
+      "loss": 0.2623,
+      "step": 26415
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.846898195669491e-07,
+      "loss": 0.3321,
+      "step": 26416
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8438202317037987e-07,
+      "loss": 0.3031,
+      "step": 26417
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.840743475465136e-07,
+      "loss": 0.3483,
+      "step": 26418
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8376679269920946e-07,
+      "loss": 0.3156,
+      "step": 26419
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.834593586323332e-07,
+      "loss": 0.3055,
+      "step": 26420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8315204534974726e-07,
+      "loss": 0.3417,
+      "step": 26421
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.828448528553108e-07,
+      "loss": 0.303,
+      "step": 26422
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8253778115287963e-07,
+      "loss": 0.3172,
+      "step": 26423
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.82230830246314e-07,
+      "loss": 0.2805,
+      "step": 26424
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.819240001394675e-07,
+      "loss": 0.3362,
+      "step": 26425
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.816172908361926e-07,
+      "loss": 0.3083,
+      "step": 26426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8131070234034617e-07,
+      "loss": 0.323,
+      "step": 26427
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8100423465577406e-07,
+      "loss": 0.3092,
+      "step": 26428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8069788778632655e-07,
+      "loss": 0.2958,
+      "step": 26429
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8039166173585497e-07,
+      "loss": 0.3024,
+      "step": 26430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.800855565082007e-07,
+      "loss": 0.3184,
+      "step": 26431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.797795721072117e-07,
+      "loss": 0.2978,
+      "step": 26432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.794737085367306e-07,
+      "loss": 0.2821,
+      "step": 26433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7916796580059755e-07,
+      "loss": 0.3103,
+      "step": 26434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7886234390265286e-07,
+      "loss": 0.3333,
+      "step": 26435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7855684284673675e-07,
+      "loss": 0.3383,
+      "step": 26436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7825146263668624e-07,
+      "loss": 0.3023,
+      "step": 26437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7794620327633593e-07,
+      "loss": 0.3118,
+      "step": 26438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7764106476951946e-07,
+      "loss": 0.3056,
+      "step": 26439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7733604712007155e-07,
+      "loss": 0.3257,
+      "step": 26440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7703115033182026e-07,
+      "loss": 0.291,
+      "step": 26441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7672637440859804e-07,
+      "loss": 0.3108,
+      "step": 26442
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.764217193542319e-07,
+      "loss": 0.3157,
+      "step": 26443
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7611718517254645e-07,
+      "loss": 0.2603,
+      "step": 26444
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7581277186736764e-07,
+      "loss": 0.269,
+      "step": 26445
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7550847944252124e-07,
+      "loss": 0.2964,
+      "step": 26446
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.752043079018264e-07,
+      "loss": 0.3059,
+      "step": 26447
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7490025724910337e-07,
+      "loss": 0.3045,
+      "step": 26448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.745963274881748e-07,
+      "loss": 0.3239,
+      "step": 26449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.74292518622853e-07,
+      "loss": 0.288,
+      "step": 26450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7398883065695724e-07,
+      "loss": 0.3059,
+      "step": 26451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.736852635943022e-07,
+      "loss": 0.3131,
+      "step": 26452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7338181743869827e-07,
+      "loss": 0.3128,
+      "step": 26453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.730784921939579e-07,
+      "loss": 0.3127,
+      "step": 26454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7277528786389127e-07,
+      "loss": 0.2863,
+      "step": 26455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7247220445230437e-07,
+      "loss": 0.3134,
+      "step": 26456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.721692419630074e-07,
+      "loss": 0.2729,
+      "step": 26457
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.71866400399804e-07,
+      "loss": 0.3102,
+      "step": 26458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7156367976649677e-07,
+      "loss": 0.3119,
+      "step": 26459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.712610800668892e-07,
+      "loss": 0.283,
+      "step": 26460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7095860130478277e-07,
+      "loss": 0.3053,
+      "step": 26461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7065624348397444e-07,
+      "loss": 0.3199,
+      "step": 26462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.703540066082645e-07,
+      "loss": 0.3086,
+      "step": 26463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.700518906814465e-07,
+      "loss": 0.2944,
+      "step": 26464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.697498957073165e-07,
+      "loss": 0.3117,
+      "step": 26465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6944802168966676e-07,
+      "loss": 0.3141,
+      "step": 26466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6914626863229e-07,
+      "loss": 0.3117,
+      "step": 26467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6884463653897415e-07,
+      "loss": 0.3179,
+      "step": 26468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.685431254135097e-07,
+      "loss": 0.2959,
+      "step": 26469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6824173525968343e-07,
+      "loss": 0.3235,
+      "step": 26470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.679404660812802e-07,
+      "loss": 0.3266,
+      "step": 26471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6763931788208363e-07,
+      "loss": 0.3104,
+      "step": 26472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.673382906658773e-07,
+      "loss": 0.2536,
+      "step": 26473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6703738443644275e-07,
+      "loss": 0.338,
+      "step": 26474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.667365991975558e-07,
+      "loss": 0.2998,
+      "step": 26475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.664359349529967e-07,
+      "loss": 0.3272,
+      "step": 26476
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6613539170654356e-07,
+      "loss": 0.3119,
+      "step": 26477
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.658349694619689e-07,
+      "loss": 0.2768,
+      "step": 26478
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.655346682230454e-07,
+      "loss": 0.2499,
+      "step": 26479
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6523448799354764e-07,
+      "loss": 0.3041,
+      "step": 26480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6493442877724383e-07,
+      "loss": 0.3194,
+      "step": 26481
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6463449057790203e-07,
+      "loss": 0.3011,
+      "step": 26482
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6433467339929254e-07,
+      "loss": 0.3139,
+      "step": 26483
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.64034977245179e-07,
+      "loss": 0.2973,
+      "step": 26484
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6373540211932513e-07,
+      "loss": 0.3346,
+      "step": 26485
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634359480254945e-07,
+      "loss": 0.31,
+      "step": 26486
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.631366149674476e-07,
+      "loss": 0.3103,
+      "step": 26487
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6283740294894455e-07,
+      "loss": 0.3077,
+      "step": 26488
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6253831197374466e-07,
+      "loss": 0.3343,
+      "step": 26489
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6223934204560165e-07,
+      "loss": 0.2928,
+      "step": 26490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.619404931682724e-07,
+      "loss": 0.3151,
+      "step": 26491
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.616417653455118e-07,
+      "loss": 0.3299,
+      "step": 26492
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6134315858106894e-07,
+      "loss": 0.2706,
+      "step": 26493
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.610446728786976e-07,
+      "loss": 0.3029,
+      "step": 26494
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6074630824214364e-07,
+      "loss": 0.3178,
+      "step": 26495
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6044806467515624e-07,
+      "loss": 0.3106,
+      "step": 26496
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6014994218148137e-07,
+      "loss": 0.2996,
+      "step": 26497
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5985194076486263e-07,
+      "loss": 0.308,
+      "step": 26498
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.595540604290437e-07,
+      "loss": 0.3057,
+      "step": 26499
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.59256301177765e-07,
+      "loss": 0.33,
+      "step": 26500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.589586630147679e-07,
+      "loss": 0.2825,
+      "step": 26501
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5866114594378945e-07,
+      "loss": 0.3038,
+      "step": 26502
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.583637499685666e-07,
+      "loss": 0.3076,
+      "step": 26503
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5806647509283645e-07,
+      "loss": 0.3341,
+      "step": 26504
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5776932132033037e-07,
+      "loss": 0.2823,
+      "step": 26505
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.574722886547821e-07,
+      "loss": 0.3104,
+      "step": 26506
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5717537709992314e-07,
+      "loss": 0.3172,
+      "step": 26507
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5687858665947927e-07,
+      "loss": 0.258,
+      "step": 26508
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5658191733718095e-07,
+      "loss": 0.3198,
+      "step": 26509
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5628536913675406e-07,
+      "loss": 0.3345,
+      "step": 26510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.559889420619234e-07,
+      "loss": 0.3023,
+      "step": 26511
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5569263611641034e-07,
+      "loss": 0.2597,
+      "step": 26512
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.553964513039387e-07,
+      "loss": 0.3051,
+      "step": 26513
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5510038762822885e-07,
+      "loss": 0.3074,
+      "step": 26514
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.548044450929977e-07,
+      "loss": 0.3202,
+      "step": 26515
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5450862370196346e-07,
+      "loss": 0.285,
+      "step": 26516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5421292345884093e-07,
+      "loss": 0.3205,
+      "step": 26517
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.539173443673427e-07,
+      "loss": 0.278,
+      "step": 26518
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.536218864311836e-07,
+      "loss": 0.3011,
+      "step": 26519
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5332654965407495e-07,
+      "loss": 0.2856,
+      "step": 26520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.53031334039724e-07,
+      "loss": 0.3046,
+      "step": 26521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.527362395918399e-07,
+      "loss": 0.275,
+      "step": 26522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5244126631412967e-07,
+      "loss": 0.2904,
+      "step": 26523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5214641421029705e-07,
+      "loss": 0.3263,
+      "step": 26524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5185168328404465e-07,
+      "loss": 0.3301,
+      "step": 26525
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.515570735390783e-07,
+      "loss": 0.3178,
+      "step": 26526
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5126258497909405e-07,
+      "loss": 0.2976,
+      "step": 26527
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.509682176077911e-07,
+      "loss": 0.2822,
+      "step": 26528
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5067397142886984e-07,
+      "loss": 0.3061,
+      "step": 26529
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5037984644602173e-07,
+      "loss": 0.323,
+      "step": 26530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.500858426629439e-07,
+      "loss": 0.3001,
+      "step": 26531
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4979196008332885e-07,
+      "loss": 0.3262,
+      "step": 26532
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4949819871086586e-07,
+      "loss": 0.3414,
+      "step": 26533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.492045585492465e-07,
+      "loss": 0.3113,
+      "step": 26534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4891103960216e-07,
+      "loss": 0.3008,
+      "step": 26535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4861764187329004e-07,
+      "loss": 0.3236,
+      "step": 26536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4832436536632374e-07,
+      "loss": 0.3075,
+      "step": 26537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4803121008494366e-07,
+      "loss": 0.3221,
+      "step": 26538
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4773817603283133e-07,
+      "loss": 0.3006,
+      "step": 26539
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.474452632136682e-07,
+      "loss": 0.3097,
+      "step": 26540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4715247163113473e-07,
+      "loss": 0.2858,
+      "step": 26541
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.468598012889057e-07,
+      "loss": 0.3208,
+      "step": 26542
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.465672521906571e-07,
+      "loss": 0.3104,
+      "step": 26543
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4627482434006707e-07,
+      "loss": 0.2861,
+      "step": 26544
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4598251774080385e-07,
+      "loss": 0.3186,
+      "step": 26545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.45690332396541e-07,
+      "loss": 0.3018,
+      "step": 26546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4539826831094916e-07,
+      "loss": 0.2964,
+      "step": 26547
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.451063254876963e-07,
+      "loss": 0.327,
+      "step": 26548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4481450393044733e-07,
+      "loss": 0.3078,
+      "step": 26549
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.445228036428694e-07,
+      "loss": 0.3208,
+      "step": 26550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4423122462862614e-07,
+      "loss": 0.294,
+      "step": 26551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.439397668913791e-07,
+      "loss": 0.3284,
+      "step": 26552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.436484304347887e-07,
+      "loss": 0.3256,
+      "step": 26553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.433572152625164e-07,
+      "loss": 0.3197,
+      "step": 26554
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.430661213782172e-07,
+      "loss": 0.3021,
+      "step": 26555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4277514878554797e-07,
+      "loss": 0.3146,
+      "step": 26556
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.424842974881659e-07,
+      "loss": 0.3193,
+      "step": 26557
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4219356748972146e-07,
+      "loss": 0.3352,
+      "step": 26558
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4190295879386495e-07,
+      "loss": 0.3163,
+      "step": 26559
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4161247140424903e-07,
+      "loss": 0.33,
+      "step": 26560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4132210532452193e-07,
+      "loss": 0.3071,
+      "step": 26561
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.410318605583285e-07,
+      "loss": 0.315,
+      "step": 26562
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.3321,
+      "step": 26563
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.404517349811309e-07,
+      "loss": 0.2957,
+      "step": 26564
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.40161854177411e-07,
+      "loss": 0.3125,
+      "step": 26565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3987209470179974e-07,
+      "loss": 0.2825,
+      "step": 26566
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.395824565579353e-07,
+      "loss": 0.3019,
+      "step": 26567
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.392929397494571e-07,
+      "loss": 0.3021,
+      "step": 26568
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.390035442799999e-07,
+      "loss": 0.3164,
+      "step": 26569
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.387142701532009e-07,
+      "loss": 0.2882,
+      "step": 26570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.384251173726893e-07,
+      "loss": 0.331,
+      "step": 26571
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3813608594210236e-07,
+      "loss": 0.3344,
+      "step": 26572
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.378471758650648e-07,
+      "loss": 0.3036,
+      "step": 26573
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3755838714520817e-07,
+      "loss": 0.3089,
+      "step": 26574
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3726971978615964e-07,
+      "loss": 0.2891,
+      "step": 26575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3698117379154515e-07,
+      "loss": 0.2833,
+      "step": 26576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.366927491649874e-07,
+      "loss": 0.257,
+      "step": 26577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.364044459101101e-07,
+      "loss": 0.311,
+      "step": 26578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3611626403053487e-07,
+      "loss": 0.3113,
+      "step": 26579
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.358282035298799e-07,
+      "loss": 0.3351,
+      "step": 26580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.355402644117656e-07,
+      "loss": 0.277,
+      "step": 26581
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3525244667980463e-07,
+      "loss": 0.322,
+      "step": 26582
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.349647503376152e-07,
+      "loss": 0.3129,
+      "step": 26583
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3467717538880893e-07,
+      "loss": 0.2766,
+      "step": 26584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.343897218369996e-07,
+      "loss": 0.316,
+      "step": 26585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3410238968579646e-07,
+      "loss": 0.337,
+      "step": 26586
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.338151789388078e-07,
+      "loss": 0.2989,
+      "step": 26587
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.33528089599644e-07,
+      "loss": 0.3026,
+      "step": 26588
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.332411216719067e-07,
+      "loss": 0.3135,
+      "step": 26589
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.329542751592041e-07,
+      "loss": 0.3409,
+      "step": 26590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.326675500651366e-07,
+      "loss": 0.3133,
+      "step": 26591
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.323809463933059e-07,
+      "loss": 0.3054,
+      "step": 26592
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3209446414731115e-07,
+      "loss": 0.3113,
+      "step": 26593
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.31808103330753e-07,
+      "loss": 0.3265,
+      "step": 26594
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3152186394722506e-07,
+      "loss": 0.3019,
+      "step": 26595
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.312357460003235e-07,
+      "loss": 0.2963,
+      "step": 26596
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.309497494936442e-07,
+      "loss": 0.3026,
+      "step": 26597
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3066387443077665e-07,
+      "loss": 0.2913,
+      "step": 26598
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.303781208153123e-07,
+      "loss": 0.3299,
+      "step": 26599
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.300924886508416e-07,
+      "loss": 0.2939,
+      "step": 26600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2980697794095075e-07,
+      "loss": 0.3219,
+      "step": 26601
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2952158868922445e-07,
+      "loss": 0.3017,
+      "step": 26602
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2923632089924993e-07,
+      "loss": 0.3222,
+      "step": 26603
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.289511745746077e-07,
+      "loss": 0.2854,
+      "step": 26604
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2866614971888035e-07,
+      "loss": 0.3013,
+      "step": 26605
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.283812463356484e-07,
+      "loss": 0.3247,
+      "step": 26606
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2809646442848895e-07,
+      "loss": 0.2954,
+      "step": 26607
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.278118040009792e-07,
+      "loss": 0.2775,
+      "step": 26608
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.275272650566963e-07,
+      "loss": 0.2939,
+      "step": 26609
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2724284759921177e-07,
+      "loss": 0.3197,
+      "step": 26610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2695855163210056e-07,
+      "loss": 0.323,
+      "step": 26611
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.266743771589309e-07,
+      "loss": 0.3163,
+      "step": 26612
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.263903241832711e-07,
+      "loss": 0.2633,
+      "step": 26613
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.261063927086916e-07,
+      "loss": 0.3031,
+      "step": 26614
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2582258273875846e-07,
+      "loss": 0.2851,
+      "step": 26615
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2553889427703435e-07,
+      "loss": 0.2839,
+      "step": 26616
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.252553273270842e-07,
+      "loss": 0.3053,
+      "step": 26617
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2497188189246745e-07,
+      "loss": 0.3212,
+      "step": 26618
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.246885579767478e-07,
+      "loss": 0.2985,
+      "step": 26619
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.244053555834814e-07,
+      "loss": 0.3211,
+      "step": 26620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.241222747162254e-07,
+      "loss": 0.3249,
+      "step": 26621
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2383931537853577e-07,
+      "loss": 0.274,
+      "step": 26622
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2355647757396525e-07,
+      "loss": 0.288,
+      "step": 26623
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2327376130606767e-07,
+      "loss": 0.2827,
+      "step": 26624
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.229911665783947e-07,
+      "loss": 0.3102,
+      "step": 26625
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.227086933944934e-07,
+      "loss": 0.3397,
+      "step": 26626
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.224263417579132e-07,
+      "loss": 0.2699,
+      "step": 26627
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2214411167220014e-07,
+      "loss": 0.3304,
+      "step": 26628
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2186200314089923e-07,
+      "loss": 0.3202,
+      "step": 26629
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2158001616755307e-07,
+      "loss": 0.2598,
+      "step": 26630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.212981507557056e-07,
+      "loss": 0.3006,
+      "step": 26631
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2101640690889503e-07,
+      "loss": 0.2888,
+      "step": 26632
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2073478463065857e-07,
+      "loss": 0.3148,
+      "step": 26633
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2045328392453777e-07,
+      "loss": 0.2866,
+      "step": 26634
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.201719047940632e-07,
+      "loss": 0.3205,
+      "step": 26635
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.198906472427732e-07,
+      "loss": 0.3455,
+      "step": 26636
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1960951127419924e-07,
+      "loss": 0.3148,
+      "step": 26637
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1932849689187083e-07,
+      "loss": 0.2934,
+      "step": 26638
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1904760409931846e-07,
+      "loss": 0.3346,
+      "step": 26639
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.187668329000715e-07,
+      "loss": 0.3092,
+      "step": 26640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.184861832976549e-07,
+      "loss": 0.3073,
+      "step": 26641
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1820565529559476e-07,
+      "loss": 0.3055,
+      "step": 26642
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1792524889741384e-07,
+      "loss": 0.3154,
+      "step": 26643
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.176449641066326e-07,
+      "loss": 0.2982,
+      "step": 26644
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.173648009267727e-07,
+      "loss": 0.2982,
+      "step": 26645
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.170847593613557e-07,
+      "loss": 0.2993,
+      "step": 26646
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.168048394138945e-07,
+      "loss": 0.3131,
+      "step": 26647
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1652504108790614e-07,
+      "loss": 0.3056,
+      "step": 26648
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.162453643869079e-07,
+      "loss": 0.3343,
+      "step": 26649
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.15965809314408e-07,
+      "loss": 0.3434,
+      "step": 26650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.156863758739204e-07,
+      "loss": 0.3144,
+      "step": 26651
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.154070640689555e-07,
+      "loss": 0.3451,
+      "step": 26652
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1512787390302056e-07,
+      "loss": 0.2924,
+      "step": 26653
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.148488053796206e-07,
+      "loss": 0.3071,
+      "step": 26654
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1456985850226164e-07,
+      "loss": 0.2927,
+      "step": 26655
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1429103327444866e-07,
+      "loss": 0.336,
+      "step": 26656
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.140123296996822e-07,
+      "loss": 0.2889,
+      "step": 26657
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1373374778146395e-07,
+      "loss": 0.2883,
+      "step": 26658
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1345528752329214e-07,
+      "loss": 0.3052,
+      "step": 26659
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.13176948928664e-07,
+      "loss": 0.3258,
+      "step": 26660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.128987320010757e-07,
+      "loss": 0.3357,
+      "step": 26661
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.126206367440232e-07,
+      "loss": 0.2738,
+      "step": 26662
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.123426631609983e-07,
+      "loss": 0.3381,
+      "step": 26663
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1206481125549027e-07,
+      "loss": 0.3145,
+      "step": 26664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.117870810309931e-07,
+      "loss": 0.3276,
+      "step": 26665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.115094724909906e-07,
+      "loss": 0.3349,
+      "step": 26666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.112319856389723e-07,
+      "loss": 0.3097,
+      "step": 26667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1095462047842305e-07,
+      "loss": 0.3241,
+      "step": 26668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1067737701282576e-07,
+      "loss": 0.2861,
+      "step": 26669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1040025524566417e-07,
+      "loss": 0.3089,
+      "step": 26670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1012325518041783e-07,
+      "loss": 0.26,
+      "step": 26671
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0984637682056615e-07,
+      "loss": 0.307,
+      "step": 26672
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0956962016958747e-07,
+      "loss": 0.2968,
+      "step": 26673
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.092929852309556e-07,
+      "loss": 0.3116,
+      "step": 26674
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.090164720081479e-07,
+      "loss": 0.3235,
+      "step": 26675
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.087400805046348e-07,
+      "loss": 0.3078,
+      "step": 26676
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.084638107238902e-07,
+      "loss": 0.3122,
+      "step": 26677
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.081876626693825e-07,
+      "loss": 0.2676,
+      "step": 26678
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0791163634458e-07,
+      "loss": 0.3392,
+      "step": 26679
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0763573175295103e-07,
+      "loss": 0.3077,
+      "step": 26680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.073599488979595e-07,
+      "loss": 0.3036,
+      "step": 26681
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.070842877830693e-07,
+      "loss": 0.3004,
+      "step": 26682
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0680874841174435e-07,
+      "loss": 0.2899,
+      "step": 26683
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0653333078744406e-07,
+      "loss": 0.3336,
+      "step": 26684
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0625803491362907e-07,
+      "loss": 0.2733,
+      "step": 26685
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.059828607937554e-07,
+      "loss": 0.3119,
+      "step": 26686
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.057078084312792e-07,
+      "loss": 0.273,
+      "step": 26687
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.054328778296556e-07,
+      "loss": 0.3,
+      "step": 26688
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051580689923395e-07,
+      "loss": 0.2927,
+      "step": 26689
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0488338192277924e-07,
+      "loss": 0.3312,
+      "step": 26690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.046088166244276e-07,
+      "loss": 0.2899,
+      "step": 26691
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0433437310073197e-07,
+      "loss": 0.3017,
+      "step": 26692
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.040600513551406e-07,
+      "loss": 0.3054,
+      "step": 26693
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0378585139109627e-07,
+      "loss": 0.3355,
+      "step": 26694
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.035117732120474e-07,
+      "loss": 0.3283,
+      "step": 26695
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0323781682143225e-07,
+      "loss": 0.2929,
+      "step": 26696
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0296398222269264e-07,
+      "loss": 0.3154,
+      "step": 26697
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0269026941926907e-07,
+      "loss": 0.3028,
+      "step": 26698
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0241667841459875e-07,
+      "loss": 0.3193,
+      "step": 26699
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0214320921211793e-07,
+      "loss": 0.2709,
+      "step": 26700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0186986181526044e-07,
+      "loss": 0.3153,
+      "step": 26701
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0159663622746247e-07,
+      "loss": 0.3094,
+      "step": 26702
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.013235324521535e-07,
+      "loss": 0.29,
+      "step": 26703
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.01050550492763e-07,
+      "loss": 0.3195,
+      "step": 26704
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0077769035272263e-07,
+      "loss": 0.3423,
+      "step": 26705
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.005049520354575e-07,
+      "loss": 0.291,
+      "step": 26706
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0023233554439256e-07,
+      "loss": 0.287,
+      "step": 26707
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.999598408829529e-07,
+      "loss": 0.3144,
+      "step": 26708
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.996874680545603e-07,
+      "loss": 0.3,
+      "step": 26709
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.994152170626363e-07,
+      "loss": 0.2832,
+      "step": 26710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9914308791060165e-07,
+      "loss": 0.3267,
+      "step": 26711
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.988710806018724e-07,
+      "loss": 0.3213,
+      "step": 26712
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.985991951398659e-07,
+      "loss": 0.301,
+      "step": 26713
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9832743152799714e-07,
+      "loss": 0.3191,
+      "step": 26714
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9805578976967897e-07,
+      "loss": 0.3112,
+      "step": 26715
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.977842698683231e-07,
+      "loss": 0.3095,
+      "step": 26716
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9751287182734124e-07,
+      "loss": 0.3251,
+      "step": 26717
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9724159565013955e-07,
+      "loss": 0.3246,
+      "step": 26718
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.969704413401275e-07,
+      "loss": 0.2913,
+      "step": 26719
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9669940890071136e-07,
+      "loss": 0.2894,
+      "step": 26720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9642849833529166e-07,
+      "loss": 0.3445,
+      "step": 26721
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9615770964727455e-07,
+      "loss": 0.3203,
+      "step": 26722
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9588704284006176e-07,
+      "loss": 0.2936,
+      "step": 26723
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9561649791704837e-07,
+      "loss": 0.3377,
+      "step": 26724
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9534607488163616e-07,
+      "loss": 0.3185,
+      "step": 26725
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.950757737372223e-07,
+      "loss": 0.2939,
+      "step": 26726
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.948055944871997e-07,
+      "loss": 0.3439,
+      "step": 26727
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.945355371349612e-07,
+      "loss": 0.2617,
+      "step": 26728
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9426560168390073e-07,
+      "loss": 0.3262,
+      "step": 26729
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9399578813740895e-07,
+      "loss": 0.3197,
+      "step": 26730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9372609649887196e-07,
+      "loss": 0.3046,
+      "step": 26731
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9345652677167823e-07,
+      "loss": 0.2931,
+      "step": 26732
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9318707895921614e-07,
+      "loss": 0.287,
+      "step": 26733
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9291775306486637e-07,
+      "loss": 0.3089,
+      "step": 26734
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.926485490920139e-07,
+      "loss": 0.3039,
+      "step": 26735
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.923794670440405e-07,
+      "loss": 0.3021,
+      "step": 26736
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.921105069243235e-07,
+      "loss": 0.3338,
+      "step": 26737
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9184166873624243e-07,
+      "loss": 0.3268,
+      "step": 26738
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.915729524831745e-07,
+      "loss": 0.3278,
+      "step": 26739
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.913043581684927e-07,
+      "loss": 0.3308,
+      "step": 26740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.91035885795572e-07,
+      "loss": 0.3242,
+      "step": 26741
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9076753536778526e-07,
+      "loss": 0.3193,
+      "step": 26742
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.904993068885009e-07,
+      "loss": 0.2994,
+      "step": 26743
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9023120036108854e-07,
+      "loss": 0.2667,
+      "step": 26744
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.899632157889187e-07,
+      "loss": 0.2709,
+      "step": 26745
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.896953531753521e-07,
+      "loss": 0.3166,
+      "step": 26746
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.894276125237583e-07,
+      "loss": 0.2833,
+      "step": 26747
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.891599938374956e-07,
+      "loss": 0.3107,
+      "step": 26748
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8889249711992805e-07,
+      "loss": 0.2889,
+      "step": 26749
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.886251223744141e-07,
+      "loss": 0.3185,
+      "step": 26750
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8835786960431324e-07,
+      "loss": 0.2811,
+      "step": 26751
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8809073881298054e-07,
+      "loss": 0.3154,
+      "step": 26752
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8782373000377117e-07,
+      "loss": 0.3058,
+      "step": 26753
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.875568431800424e-07,
+      "loss": 0.3011,
+      "step": 26754
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8729007834514157e-07,
+      "loss": 0.3131,
+      "step": 26755
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.870234355024215e-07,
+      "loss": 0.2921,
+      "step": 26756
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.867569146552318e-07,
+      "loss": 0.3388,
+      "step": 26757
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.864905158069198e-07,
+      "loss": 0.2916,
+      "step": 26758
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.862242389608294e-07,
+      "loss": 0.3331,
+      "step": 26759
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.85958084120308e-07,
+      "loss": 0.2987,
+      "step": 26760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8569205128869624e-07,
+      "loss": 0.2889,
+      "step": 26761
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8542614046933594e-07,
+      "loss": 0.2957,
+      "step": 26762
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.851603516655688e-07,
+      "loss": 0.2776,
+      "step": 26763
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8489468488072993e-07,
+      "loss": 0.3195,
+      "step": 26764
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.846291401181589e-07,
+      "loss": 0.3073,
+      "step": 26765
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8436371738118973e-07,
+      "loss": 0.3166,
+      "step": 26766
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.840984166731575e-07,
+      "loss": 0.299,
+      "step": 26767
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8383323799739295e-07,
+      "loss": 0.3111,
+      "step": 26768
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.835681813572266e-07,
+      "loss": 0.31,
+      "step": 26769
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.833032467559882e-07,
+      "loss": 0.3357,
+      "step": 26770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8303843419700494e-07,
+      "loss": 0.3198,
+      "step": 26771
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.827737436836031e-07,
+      "loss": 0.2827,
+      "step": 26772
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825091752191078e-07,
+      "loss": 0.3182,
+      "step": 26773
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8224472880684083e-07,
+      "loss": 0.3117,
+      "step": 26774
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8198040445012397e-07,
+      "loss": 0.3482,
+      "step": 26775
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.81716202152279e-07,
+      "loss": 0.3202,
+      "step": 26776
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.814521219166222e-07,
+      "loss": 0.2887,
+      "step": 26777
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8118816374647087e-07,
+      "loss": 0.2929,
+      "step": 26778
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.809243276451401e-07,
+      "loss": 0.319,
+      "step": 26779
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.806606136159451e-07,
+      "loss": 0.3136,
+      "step": 26780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8039702166219653e-07,
+      "loss": 0.3027,
+      "step": 26781
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8013355178720613e-07,
+      "loss": 0.314,
+      "step": 26782
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7987020399428245e-07,
+      "loss": 0.3051,
+      "step": 26783
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7960697828673277e-07,
+      "loss": 0.3211,
+      "step": 26784
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.793438746678645e-07,
+      "loss": 0.2674,
+      "step": 26785
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7908089314098164e-07,
+      "loss": 0.2728,
+      "step": 26786
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.78818033709386e-07,
+      "loss": 0.2654,
+      "step": 26787
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.785552963763827e-07,
+      "loss": 0.2726,
+      "step": 26788
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7829268114526686e-07,
+      "loss": 0.2913,
+      "step": 26789
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7803018801934147e-07,
+      "loss": 0.2809,
+      "step": 26790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.777678170019016e-07,
+      "loss": 0.2875,
+      "step": 26791
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7750556809624017e-07,
+      "loss": 0.2623,
+      "step": 26792
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.772434413056546e-07,
+      "loss": 0.2928,
+      "step": 26793
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.769814366334356e-07,
+      "loss": 0.3067,
+      "step": 26794
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.767195540828749e-07,
+      "loss": 0.3087,
+      "step": 26795
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.764577936572599e-07,
+      "loss": 0.3677,
+      "step": 26796
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.761961553598813e-07,
+      "loss": 0.3677,
+      "step": 26797
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7593463919402206e-07,
+      "loss": 0.3234,
+      "step": 26798
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7567324516296845e-07,
+      "loss": 0.2962,
+      "step": 26799
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.754119732700056e-07,
+      "loss": 0.3171,
+      "step": 26800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7515082351841195e-07,
+      "loss": 0.3002,
+      "step": 26801
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7488979591146716e-07,
+      "loss": 0.2849,
+      "step": 26802
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.746288904524519e-07,
+      "loss": 0.2688,
+      "step": 26803
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.743681071446436e-07,
+      "loss": 0.3128,
+      "step": 26804
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.741074459913151e-07,
+      "loss": 0.2977,
+      "step": 26805
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.738469069957417e-07,
+      "loss": 0.3295,
+      "step": 26806
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7358649016119733e-07,
+      "loss": 0.3023,
+      "step": 26807
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.733261954909505e-07,
+      "loss": 0.3002,
+      "step": 26808
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.730660229882709e-07,
+      "loss": 0.29,
+      "step": 26809
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7280597265642805e-07,
+      "loss": 0.2878,
+      "step": 26810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7254604449868714e-07,
+      "loss": 0.2926,
+      "step": 26811
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7228623851831224e-07,
+      "loss": 0.3064,
+      "step": 26812
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.720265547185674e-07,
+      "loss": 0.2595,
+      "step": 26813
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.717669931027123e-07,
+      "loss": 0.3275,
+      "step": 26814
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.715075536740086e-07,
+      "loss": 0.2646,
+      "step": 26815
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.712482364357172e-07,
+      "loss": 0.302,
+      "step": 26816
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.709890413910909e-07,
+      "loss": 0.319,
+      "step": 26817
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.707299685433873e-07,
+      "loss": 0.3384,
+      "step": 26818
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7047101789586026e-07,
+      "loss": 0.3043,
+      "step": 26819
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.702121894517629e-07,
+      "loss": 0.2937,
+      "step": 26820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.699534832143447e-07,
+      "loss": 0.3161,
+      "step": 26821
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6969489918685643e-07,
+      "loss": 0.2802,
+      "step": 26822
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.694364373725444e-07,
+      "loss": 0.2829,
+      "step": 26823
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.691780977746561e-07,
+      "loss": 0.3224,
+      "step": 26824
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6891988039643656e-07,
+      "loss": 0.3001,
+      "step": 26825
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.686617852411266e-07,
+      "loss": 0.3129,
+      "step": 26826
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6840381231197034e-07,
+      "loss": 0.2775,
+      "step": 26827
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.681459616122073e-07,
+      "loss": 0.3159,
+      "step": 26828
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6788823314507604e-07,
+      "loss": 0.3226,
+      "step": 26829
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6763062691381293e-07,
+      "loss": 0.3492,
+      "step": 26830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.673731429216564e-07,
+      "loss": 0.3198,
+      "step": 26831
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.671157811718372e-07,
+      "loss": 0.3343,
+      "step": 26832
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.668585416675884e-07,
+      "loss": 0.2904,
+      "step": 26833
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6660142441214177e-07,
+      "loss": 0.3494,
+      "step": 26834
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6634442940872586e-07,
+      "loss": 0.265,
+      "step": 26835
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.660875566605692e-07,
+      "loss": 0.3166,
+      "step": 26836
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6583080617089695e-07,
+      "loss": 0.3179,
+      "step": 26837
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.655741779429377e-07,
+      "loss": 0.3073,
+      "step": 26838
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6531767197991e-07,
+      "loss": 0.283,
+      "step": 26839
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6506128828503673e-07,
+      "loss": 0.2977,
+      "step": 26840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.64805026861541e-07,
+      "loss": 0.2751,
+      "step": 26841
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.64548887712639e-07,
+      "loss": 0.3137,
+      "step": 26842
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.642928708415482e-07,
+      "loss": 0.315,
+      "step": 26843
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6403697625148383e-07,
+      "loss": 0.3132,
+      "step": 26844
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6378120394566e-07,
+      "loss": 0.2874,
+      "step": 26845
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6352555392728963e-07,
+      "loss": 0.2835,
+      "step": 26846
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.632700261995835e-07,
+      "loss": 0.3272,
+      "step": 26847
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6301462076575135e-07,
+      "loss": 0.2622,
+      "step": 26848
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.627593376290005e-07,
+      "loss": 0.3314,
+      "step": 26849
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.625041767925385e-07,
+      "loss": 0.2939,
+      "step": 26850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6224913825956933e-07,
+      "loss": 0.292,
+      "step": 26851
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6199422203329607e-07,
+      "loss": 0.3278,
+      "step": 26852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6173942811692057e-07,
+      "loss": 0.3315,
+      "step": 26853
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.614847565136447e-07,
+      "loss": 0.3106,
+      "step": 26854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.612302072266637e-07,
+      "loss": 0.324,
+      "step": 26855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6097578025917836e-07,
+      "loss": 0.3201,
+      "step": 26856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6072147561438166e-07,
+      "loss": 0.3209,
+      "step": 26857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6046729329546773e-07,
+      "loss": 0.2847,
+      "step": 26858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.602132333056318e-07,
+      "loss": 0.3328,
+      "step": 26859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.599592956480623e-07,
+      "loss": 0.3145,
+      "step": 26860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5970548032594913e-07,
+      "loss": 0.3156,
+      "step": 26861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5945178734248177e-07,
+      "loss": 0.3214,
+      "step": 26862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.591982167008456e-07,
+      "loss": 0.2932,
+      "step": 26863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5894476840422456e-07,
+      "loss": 0.3378,
+      "step": 26864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5869144245580403e-07,
+      "loss": 0.3212,
+      "step": 26865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.584382388587625e-07,
+      "loss": 0.3345,
+      "step": 26866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5818515761628303e-07,
+      "loss": 0.2959,
+      "step": 26867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.579321987315453e-07,
+      "loss": 0.3052,
+      "step": 26868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5767936220772226e-07,
+      "loss": 0.3094,
+      "step": 26869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5742664804799253e-07,
+      "loss": 0.3486,
+      "step": 26870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.571740562555314e-07,
+      "loss": 0.3462,
+      "step": 26871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.569215868335084e-07,
+      "loss": 0.309,
+      "step": 26872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.566692397850956e-07,
+      "loss": 0.2926,
+      "step": 26873
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.564170151134648e-07,
+      "loss": 0.3596,
+      "step": 26874
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.561649128217814e-07,
+      "loss": 0.3173,
+      "step": 26875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.559129329132104e-07,
+      "loss": 0.3509,
+      "step": 26876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5566107539092056e-07,
+      "loss": 0.2858,
+      "step": 26877
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5540934025807265e-07,
+      "loss": 0.301,
+      "step": 26878
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5515772751782965e-07,
+      "loss": 0.3078,
+      "step": 26879
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5490623717335017e-07,
+      "loss": 0.3427,
+      "step": 26880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.546548692277961e-07,
+      "loss": 0.3047,
+      "step": 26881
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.544036236843206e-07,
+      "loss": 0.2998,
+      "step": 26882
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.541525005460821e-07,
+      "loss": 0.3011,
+      "step": 26883
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.539014998162348e-07,
+      "loss": 0.3086,
+      "step": 26884
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.536506214979306e-07,
+      "loss": 0.3221,
+      "step": 26885
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5339986559432037e-07,
+      "loss": 0.2796,
+      "step": 26886
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.531492321085538e-07,
+      "loss": 0.3123,
+      "step": 26887
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.528987210437772e-07,
+      "loss": 0.2956,
+      "step": 26888
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5264833240313926e-07,
+      "loss": 0.2759,
+      "step": 26889
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.523980661897851e-07,
+      "loss": 0.277,
+      "step": 26890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5214792240685573e-07,
+      "loss": 0.3374,
+      "step": 26891
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.518979010574951e-07,
+      "loss": 0.3086,
+      "step": 26892
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5164800214484307e-07,
+      "loss": 0.267,
+      "step": 26893
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.513982256720382e-07,
+      "loss": 0.317,
+      "step": 26894
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5114857164221797e-07,
+      "loss": 0.2838,
+      "step": 26895
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.508990400585176e-07,
+      "loss": 0.3088,
+      "step": 26896
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5064963092407024e-07,
+      "loss": 0.292,
+      "step": 26897
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5040034424200885e-07,
+      "loss": 0.2665,
+      "step": 26898
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5015118001546766e-07,
+      "loss": 0.2943,
+      "step": 26899
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.499021382475719e-07,
+      "loss": 0.3016,
+      "step": 26900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4965321894145133e-07,
+      "loss": 0.2971,
+      "step": 26901
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4940442210023343e-07,
+      "loss": 0.2948,
+      "step": 26902
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.491557477270412e-07,
+      "loss": 0.3348,
+      "step": 26903
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.489071958249989e-07,
+      "loss": 0.3169,
+      "step": 26904
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.486587663972295e-07,
+      "loss": 0.3137,
+      "step": 26905
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.484104594468517e-07,
+      "loss": 0.2964,
+      "step": 26906
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4816227497698406e-07,
+      "loss": 0.2726,
+      "step": 26907
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4791421299074525e-07,
+      "loss": 0.3372,
+      "step": 26908
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4766627349124827e-07,
+      "loss": 0.304,
+      "step": 26909
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.474184564816107e-07,
+      "loss": 0.3094,
+      "step": 26910
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.471707619649422e-07,
+      "loss": 0.3117,
+      "step": 26911
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4692318994435694e-07,
+      "loss": 0.3162,
+      "step": 26912
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.466757404229614e-07,
+      "loss": 0.2953,
+      "step": 26913
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.464284134038641e-07,
+      "loss": 0.3051,
+      "step": 26914
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.461812088901749e-07,
+      "loss": 0.2876,
+      "step": 26915
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4593412688499465e-07,
+      "loss": 0.2833,
+      "step": 26916
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.456871673914274e-07,
+      "loss": 0.3105,
+      "step": 26917
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.454403304125763e-07,
+      "loss": 0.3012,
+      "step": 26918
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4519361595154e-07,
+      "loss": 0.3283,
+      "step": 26919
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4494702401141936e-07,
+      "loss": 0.3143,
+      "step": 26920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.447005545953096e-07,
+      "loss": 0.31,
+      "step": 26921
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.444542077063072e-07,
+      "loss": 0.298,
+      "step": 26922
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.442079833475064e-07,
+      "loss": 0.3107,
+      "step": 26923
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.439618815220013e-07,
+      "loss": 0.2895,
+      "step": 26924
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4371590223287945e-07,
+      "loss": 0.2895,
+      "step": 26925
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4347004548323394e-07,
+      "loss": 0.278,
+      "step": 26926
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.432243112761501e-07,
+      "loss": 0.3261,
+      "step": 26927
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4297869961471544e-07,
+      "loss": 0.3131,
+      "step": 26928
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4273321050201414e-07,
+      "loss": 0.328,
+      "step": 26929
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4248784394113154e-07,
+      "loss": 0.2607,
+      "step": 26930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.422425999351463e-07,
+      "loss": 0.3011,
+      "step": 26931
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.419974784871415e-07,
+      "loss": 0.3002,
+      "step": 26932
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.417524796001947e-07,
+      "loss": 0.3229,
+      "step": 26933
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.415076032773833e-07,
+      "loss": 0.2856,
+      "step": 26934
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.412628495217828e-07,
+      "loss": 0.292,
+      "step": 26935
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4101821833646845e-07,
+      "loss": 0.2601,
+      "step": 26936
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4077370972451107e-07,
+      "loss": 0.3039,
+      "step": 26937
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.405293236889827e-07,
+      "loss": 0.3062,
+      "step": 26938
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.402850602329532e-07,
+      "loss": 0.3143,
+      "step": 26939
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.400409193594888e-07,
+      "loss": 0.3323,
+      "step": 26940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.397969010716572e-07,
+      "loss": 0.3015,
+      "step": 26941
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.395530053725237e-07,
+      "loss": 0.2878,
+      "step": 26942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3930923226515023e-07,
+      "loss": 0.3229,
+      "step": 26943
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.390655817526e-07,
+      "loss": 0.3129,
+      "step": 26944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.388220538379327e-07,
+      "loss": 0.3046,
+      "step": 26945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3857864852420696e-07,
+      "loss": 0.2853,
+      "step": 26946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3833536581448157e-07,
+      "loss": 0.2847,
+      "step": 26947
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3809220571180847e-07,
+      "loss": 0.3146,
+      "step": 26948
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.378491682192452e-07,
+      "loss": 0.3012,
+      "step": 26949
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3760625333984268e-07,
+      "loss": 0.3153,
+      "step": 26950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3736346107665287e-07,
+      "loss": 0.3154,
+      "step": 26951
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.371207914327245e-07,
+      "loss": 0.3321,
+      "step": 26952
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3687824441110508e-07,
+      "loss": 0.2939,
+      "step": 26953
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.366358200148422e-07,
+      "loss": 0.3223,
+      "step": 26954
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.363935182469812e-07,
+      "loss": 0.325,
+      "step": 26955
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.36151339110563e-07,
+      "loss": 0.3139,
+      "step": 26956
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3590928260863178e-07,
+      "loss": 0.2756,
+      "step": 26957
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3566734874422626e-07,
+      "loss": 0.3094,
+      "step": 26958
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3542553752038623e-07,
+      "loss": 0.2935,
+      "step": 26959
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3518384894014813e-07,
+      "loss": 0.2924,
+      "step": 26960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3494228300654842e-07,
+      "loss": 0.3527,
+      "step": 26961
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3470083972261914e-07,
+      "loss": 0.3199,
+      "step": 26962
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3445951909139454e-07,
+      "loss": 0.3,
+      "step": 26963
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.342183211159066e-07,
+      "loss": 0.3253,
+      "step": 26964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.339772457991818e-07,
+      "loss": 0.3247,
+      "step": 26965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3373629314424995e-07,
+      "loss": 0.311,
+      "step": 26966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3349546315413863e-07,
+      "loss": 0.3389,
+      "step": 26967
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3325475583186986e-07,
+      "loss": 0.3263,
+      "step": 26968
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3301417118046898e-07,
+      "loss": 0.3142,
+      "step": 26969
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3277370920295806e-07,
+      "loss": 0.314,
+      "step": 26970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3253336990235465e-07,
+      "loss": 0.3097,
+      "step": 26971
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3229315328167967e-07,
+      "loss": 0.3244,
+      "step": 26972
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3205305934395072e-07,
+      "loss": 0.3288,
+      "step": 26973
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3181308809218094e-07,
+      "loss": 0.3203,
+      "step": 26974
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3157323952938683e-07,
+      "loss": 0.2883,
+      "step": 26975
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3133351365858038e-07,
+      "loss": 0.3041,
+      "step": 26976
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.310939104827703e-07,
+      "loss": 0.3186,
+      "step": 26977
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3085443000496865e-07,
+      "loss": 0.3407,
+      "step": 26978
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3061507222818303e-07,
+      "loss": 0.2925,
+      "step": 26979
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.303758371554199e-07,
+      "loss": 0.3219,
+      "step": 26980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.301367247896824e-07,
+      "loss": 0.3222,
+      "step": 26981
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2989773513397483e-07,
+      "loss": 0.3041,
+      "step": 26982
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2965886819129924e-07,
+      "loss": 0.3323,
+      "step": 26983
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.294201239646543e-07,
+      "loss": 0.299,
+      "step": 26984
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2918150245703985e-07,
+      "loss": 0.305,
+      "step": 26985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.289430036714546e-07,
+      "loss": 0.2757,
+      "step": 26986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2870462761089173e-07,
+      "loss": 0.307,
+      "step": 26987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.284663742783455e-07,
+      "loss": 0.2785,
+      "step": 26988
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2822824367681017e-07,
+      "loss": 0.3173,
+      "step": 26989
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.279902358092745e-07,
+      "loss": 0.3152,
+      "step": 26990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2775235067872825e-07,
+      "loss": 0.3359,
+      "step": 26991
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2751458828816132e-07,
+      "loss": 0.3153,
+      "step": 26992
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2727694864055684e-07,
+      "loss": 0.3103,
+      "step": 26993
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.270394317389013e-07,
+      "loss": 0.3271,
+      "step": 26994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2680203758617904e-07,
+      "loss": 0.3023,
+      "step": 26995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2656476618536981e-07,
+      "loss": 0.3103,
+      "step": 26996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2632761753945353e-07,
+      "loss": 0.3196,
+      "step": 26997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2609059165141223e-07,
+      "loss": 0.3124,
+      "step": 26998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2585368852421796e-07,
+      "loss": 0.3171,
+      "step": 26999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.256169081608506e-07,
+      "loss": 0.2948,
+      "step": 27000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2538025056428216e-07,
+      "loss": 0.2933,
+      "step": 27001
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.251437157374847e-07,
+      "loss": 0.2806,
+      "step": 27002
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2490730368342928e-07,
+      "loss": 0.3184,
+      "step": 27003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2467101440508566e-07,
+      "loss": 0.2761,
+      "step": 27004
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2443484790542148e-07,
+      "loss": 0.295,
+      "step": 27005
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2419880418740325e-07,
+      "loss": 0.3064,
+      "step": 27006
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.239628832539964e-07,
+      "loss": 0.3163,
+      "step": 27007
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.237270851081619e-07,
+      "loss": 0.2965,
+      "step": 27008
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2349140975286287e-07,
+      "loss": 0.2519,
+      "step": 27009
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2325585719106036e-07,
+      "loss": 0.2952,
+      "step": 27010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2302042742571194e-07,
+      "loss": 0.323,
+      "step": 27011
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2278512045977417e-07,
+      "loss": 0.3475,
+      "step": 27012
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2254993629620247e-07,
+      "loss": 0.3093,
+      "step": 27013
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2231487493795112e-07,
+      "loss": 0.3047,
+      "step": 27014
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.220799363879722e-07,
+      "loss": 0.3077,
+      "step": 27015
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2184512064921782e-07,
+      "loss": 0.2713,
+      "step": 27016
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.216104277246356e-07,
+      "loss": 0.2781,
+      "step": 27017
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2137585761717427e-07,
+      "loss": 0.2721,
+      "step": 27018
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2114141032978043e-07,
+      "loss": 0.3072,
+      "step": 27019
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2090708586539723e-07,
+      "loss": 0.2729,
+      "step": 27020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2067288422697008e-07,
+      "loss": 0.3261,
+      "step": 27021
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2043880541743778e-07,
+      "loss": 0.3071,
+      "step": 27022
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2020484943974352e-07,
+      "loss": 0.3087,
+      "step": 27023
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.199710162968216e-07,
+      "loss": 0.3084,
+      "step": 27024
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1973730599161303e-07,
+      "loss": 0.2804,
+      "step": 27025
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.195037185270521e-07,
+      "loss": 0.2695,
+      "step": 27026
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1927025390607093e-07,
+      "loss": 0.3151,
+      "step": 27027
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1903691213160383e-07,
+      "loss": 0.3007,
+      "step": 27028
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1880369320658067e-07,
+      "loss": 0.2953,
+      "step": 27029
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1857059713393137e-07,
+      "loss": 0.2525,
+      "step": 27030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1833762391658353e-07,
+      "loss": 0.3362,
+      "step": 27031
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1810477355746261e-07,
+      "loss": 0.3047,
+      "step": 27032
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1787204605949296e-07,
+      "loss": 0.3106,
+      "step": 27033
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1763944142559778e-07,
+      "loss": 0.3255,
+      "step": 27034
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.174069596587003e-07,
+      "loss": 0.2928,
+      "step": 27035
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1717460076171925e-07,
+      "loss": 0.3041,
+      "step": 27036
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1694236473757234e-07,
+      "loss": 0.2511,
+      "step": 27037
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1671025158917835e-07,
+      "loss": 0.2895,
+      "step": 27038
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.164782613194505e-07,
+      "loss": 0.29,
+      "step": 27039
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1624639393130308e-07,
+      "loss": 0.3272,
+      "step": 27040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.160146494276505e-07,
+      "loss": 0.3177,
+      "step": 27041
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1578302781140149e-07,
+      "loss": 0.3223,
+      "step": 27042
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1555152908546484e-07,
+      "loss": 0.2815,
+      "step": 27043
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1532015325275047e-07,
+      "loss": 0.3087,
+      "step": 27044
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.150889003161616e-07,
+      "loss": 0.3094,
+      "step": 27045
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1485777027860366e-07,
+      "loss": 0.269,
+      "step": 27046
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1462676314298215e-07,
+      "loss": 0.3194,
+      "step": 27047
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.143958789121947e-07,
+      "loss": 0.3077,
+      "step": 27048
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1416511758914348e-07,
+      "loss": 0.2949,
+      "step": 27049
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1393447917672726e-07,
+      "loss": 0.2944,
+      "step": 27050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1370396367784152e-07,
+      "loss": 0.2947,
+      "step": 27051
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.134735710953828e-07,
+      "loss": 0.2877,
+      "step": 27052
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1324330143224325e-07,
+      "loss": 0.3188,
+      "step": 27053
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.130131546913161e-07,
+      "loss": 0.3182,
+      "step": 27054
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127831308754924e-07,
+      "loss": 0.3125,
+      "step": 27055
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1255322998765981e-07,
+      "loss": 0.3248,
+      "step": 27056
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1232345203070715e-07,
+      "loss": 0.3389,
+      "step": 27057
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1209379700751985e-07,
+      "loss": 0.3323,
+      "step": 27058
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1186426492098234e-07,
+      "loss": 0.326,
+      "step": 27059
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1163485577398001e-07,
+      "loss": 0.3286,
+      "step": 27060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.114055695693895e-07,
+      "loss": 0.3286,
+      "step": 27061
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1117640631009516e-07,
+      "loss": 0.2986,
+      "step": 27062
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1094736599897246e-07,
+      "loss": 0.3249,
+      "step": 27063
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1071844863889912e-07,
+      "loss": 0.2897,
+      "step": 27064
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.104896542327506e-07,
+      "loss": 0.3129,
+      "step": 27065
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1026098278340012e-07,
+      "loss": 0.3227,
+      "step": 27066
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.100324342937199e-07,
+      "loss": 0.3154,
+      "step": 27067
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.098040087665809e-07,
+      "loss": 0.3191,
+      "step": 27068
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0957570620485202e-07,
+      "loss": 0.3115,
+      "step": 27069
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.093475266113998e-07,
+      "loss": 0.2663,
+      "step": 27070
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0911946998909082e-07,
+      "loss": 0.3094,
+      "step": 27071
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0889153634079063e-07,
+      "loss": 0.2806,
+      "step": 27072
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0866372566936022e-07,
+      "loss": 0.2916,
+      "step": 27073
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0843603797766287e-07,
+      "loss": 0.2985,
+      "step": 27074
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.082084732685574e-07,
+      "loss": 0.2554,
+      "step": 27075
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0798103154490156e-07,
+      "loss": 0.2824,
+      "step": 27076
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0775371280955082e-07,
+      "loss": 0.2835,
+      "step": 27077
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0752651706536397e-07,
+      "loss": 0.3024,
+      "step": 27078
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.07299444315191e-07,
+      "loss": 0.2609,
+      "step": 27079
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0707249456188628e-07,
+      "loss": 0.3071,
+      "step": 27080
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0684566780829973e-07,
+      "loss": 0.3222,
+      "step": 27081
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.066189640572791e-07,
+      "loss": 0.2956,
+      "step": 27082
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.063923833116732e-07,
+      "loss": 0.311,
+      "step": 27083
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0616592557432758e-07,
+      "loss": 0.2587,
+      "step": 27084
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0593959084808769e-07,
+      "loss": 0.2662,
+      "step": 27085
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0571337913579347e-07,
+      "loss": 0.3245,
+      "step": 27086
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0548729044028827e-07,
+      "loss": 0.3394,
+      "step": 27087
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0526132476441085e-07,
+      "loss": 0.2995,
+      "step": 27088
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.05035482110999e-07,
+      "loss": 0.3128,
+      "step": 27089
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0480976248289153e-07,
+      "loss": 0.2814,
+      "step": 27090
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0458416588292175e-07,
+      "loss": 0.2625,
+      "step": 27091
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.043586923139229e-07,
+      "loss": 0.32,
+      "step": 27092
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0413334177872724e-07,
+      "loss": 0.2974,
+      "step": 27093
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0390811428016689e-07,
+      "loss": 0.3314,
+      "step": 27094
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0368300982106847e-07,
+      "loss": 0.2862,
+      "step": 27095
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0345802840425978e-07,
+      "loss": 0.2989,
+      "step": 27096
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0323317003256848e-07,
+      "loss": 0.3345,
+      "step": 27097
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.030084347088157e-07,
+      "loss": 0.3372,
+      "step": 27098
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0278382243582472e-07,
+      "loss": 0.3235,
+      "step": 27099
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0255933321641996e-07,
+      "loss": 0.3248,
+      "step": 27100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0233496705341692e-07,
+      "loss": 0.3127,
+      "step": 27101
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0211072394963561e-07,
+      "loss": 0.3531,
+      "step": 27102
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0188660390789372e-07,
+      "loss": 0.3079,
+      "step": 27103
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0166260693100348e-07,
+      "loss": 0.2679,
+      "step": 27104
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.014387330217804e-07,
+      "loss": 0.302,
+      "step": 27105
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.012149821830367e-07,
+      "loss": 0.3463,
+      "step": 27106
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.00991354417579e-07,
+      "loss": 0.2893,
+      "step": 27107
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.007678497282206e-07,
+      "loss": 0.2916,
+      "step": 27108
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0054446811776596e-07,
+      "loss": 0.341,
+      "step": 27109
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0032120958902168e-07,
+      "loss": 0.3279,
+      "step": 27110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.000980741447922e-07,
+      "loss": 0.2975,
+      "step": 27111
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9987506178787973e-07,
+      "loss": 0.3074,
+      "step": 27112
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.996521725210854e-07,
+      "loss": 0.3137,
+      "step": 27113
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9942940634720799e-07,
+      "loss": 0.3246,
+      "step": 27114
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9920676326904642e-07,
+      "loss": 0.3306,
+      "step": 27115
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9898424328939848e-07,
+      "loss": 0.3059,
+      "step": 27116
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9876184641105523e-07,
+      "loss": 0.3377,
+      "step": 27117
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9853957263681333e-07,
+      "loss": 0.2957,
+      "step": 27118
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.983174219694617e-07,
+      "loss": 0.2512,
+      "step": 27119
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.980953944117936e-07,
+      "loss": 0.3049,
+      "step": 27120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9787348996659682e-07,
+      "loss": 0.3219,
+      "step": 27121
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9765170863665694e-07,
+      "loss": 0.2784,
+      "step": 27122
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9743005042475948e-07,
+      "loss": 0.2814,
+      "step": 27123
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9720851533369222e-07,
+      "loss": 0.3145,
+      "step": 27124
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9698710336623295e-07,
+      "loss": 0.2983,
+      "step": 27125
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9676581452516608e-07,
+      "loss": 0.2834,
+      "step": 27126
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.965446488132683e-07,
+      "loss": 0.3139,
+      "step": 27127
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9632360623332068e-07,
+      "loss": 0.3113,
+      "step": 27128
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.961026867880955e-07,
+      "loss": 0.3301,
+      "step": 27129
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9588189048037055e-07,
+      "loss": 0.3014,
+      "step": 27130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.956612173129191e-07,
+      "loss": 0.2856,
+      "step": 27131
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9544066728851008e-07,
+      "loss": 0.3108,
+      "step": 27132
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9522024040991682e-07,
+      "loss": 0.3318,
+      "step": 27133
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9499993667990603e-07,
+      "loss": 0.2884,
+      "step": 27134
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9477975610124433e-07,
+      "loss": 0.303,
+      "step": 27135
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9455969867669955e-07,
+      "loss": 0.2793,
+      "step": 27136
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.943397644090328e-07,
+      "loss": 0.2999,
+      "step": 27137
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9411995330100963e-07,
+      "loss": 0.3032,
+      "step": 27138
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9390026535538675e-07,
+      "loss": 0.2771,
+      "step": 27139
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.936807005749275e-07,
+      "loss": 0.3219,
+      "step": 27140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9346125896238742e-07,
+      "loss": 0.3355,
+      "step": 27141
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9324194052052215e-07,
+      "loss": 0.2747,
+      "step": 27142
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.930227452520883e-07,
+      "loss": 0.3225,
+      "step": 27143
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9280367315983818e-07,
+      "loss": 0.3252,
+      "step": 27144
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9258472424652174e-07,
+      "loss": 0.3321,
+      "step": 27145
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.923658985148924e-07,
+      "loss": 0.3117,
+      "step": 27146
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.3085,
+      "step": 27147
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9192861660767947e-07,
+      "loss": 0.3108,
+      "step": 27148
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.917101604375904e-07,
+      "loss": 0.2907,
+      "step": 27149
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.914918274601696e-07,
+      "loss": 0.2943,
+      "step": 27150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9127361767816045e-07,
+      "loss": 0.2669,
+      "step": 27151
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.910555310943052e-07,
+      "loss": 0.2915,
+      "step": 27152
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9083756771134054e-07,
+      "loss": 0.3448,
+      "step": 27153
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061972753200542e-07,
+      "loss": 0.3173,
+      "step": 27154
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.904020105590376e-07,
+      "loss": 0.2719,
+      "step": 27155
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.901844167951683e-07,
+      "loss": 0.3002,
+      "step": 27156
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8996694624313418e-07,
+      "loss": 0.3,
+      "step": 27157
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.897495989056619e-07,
+      "loss": 0.2841,
+      "step": 27158
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8953237478548604e-07,
+      "loss": 0.2762,
+      "step": 27159
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.893152738853321e-07,
+      "loss": 0.3353,
+      "step": 27160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8909829620792907e-07,
+      "loss": 0.2825,
+      "step": 27161
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8888144175599921e-07,
+      "loss": 0.334,
+      "step": 27162
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.886647105322681e-07,
+      "loss": 0.2647,
+      "step": 27163
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.884481025394591e-07,
+      "loss": 0.3258,
+      "step": 27164
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.882316177802901e-07,
+      "loss": 0.3365,
+      "step": 27165
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8801525625748217e-07,
+      "loss": 0.3091,
+      "step": 27166
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.877990179737521e-07,
+      "loss": 0.2939,
+      "step": 27167
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8758290293181547e-07,
+      "loss": 0.3133,
+      "step": 27168
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8736691113438787e-07,
+      "loss": 0.2801,
+      "step": 27169
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.871510425841805e-07,
+      "loss": 0.323,
+      "step": 27170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.869352972839067e-07,
+      "loss": 0.3214,
+      "step": 27171
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8671967523627431e-07,
+      "loss": 0.3144,
+      "step": 27172
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8650417644399232e-07,
+      "loss": 0.2932,
+      "step": 27173
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8628880090976852e-07,
+      "loss": 0.2725,
+      "step": 27174
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8607354863630632e-07,
+      "loss": 0.3405,
+      "step": 27175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8585841962630914e-07,
+      "loss": 0.2767,
+      "step": 27176
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8564341388248141e-07,
+      "loss": 0.3139,
+      "step": 27177
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8542853140752103e-07,
+      "loss": 0.3107,
+      "step": 27178
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8521377220412917e-07,
+      "loss": 0.3241,
+      "step": 27179
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8499913627500144e-07,
+      "loss": 0.3267,
+      "step": 27180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8478462362283455e-07,
+      "loss": 0.2985,
+      "step": 27181
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8457023425032083e-07,
+      "loss": 0.2613,
+      "step": 27182
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.84355968160157e-07,
+      "loss": 0.294,
+      "step": 27183
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.841418253550298e-07,
+      "loss": 0.3057,
+      "step": 27184
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8392780583763148e-07,
+      "loss": 0.279,
+      "step": 27185
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8371390961065105e-07,
+      "loss": 0.2812,
+      "step": 27186
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.83500136676773e-07,
+      "loss": 0.3183,
+      "step": 27187
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8328648703868191e-07,
+      "loss": 0.3011,
+      "step": 27188
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8307296069906445e-07,
+      "loss": 0.2774,
+      "step": 27189
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8285955766059958e-07,
+      "loss": 0.3203,
+      "step": 27190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8264627792596745e-07,
+      "loss": 0.2906,
+      "step": 27191
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.824331214978492e-07,
+      "loss": 0.2544,
+      "step": 27192
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8222008837891937e-07,
+      "loss": 0.2823,
+      "step": 27193
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.820071785718558e-07,
+      "loss": 0.2955,
+      "step": 27194
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8179439207933304e-07,
+      "loss": 0.3092,
+      "step": 27195
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8158172890402115e-07,
+      "loss": 0.3289,
+      "step": 27196
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8136918904859247e-07,
+      "loss": 0.3048,
+      "step": 27197
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8115677251571816e-07,
+      "loss": 0.293,
+      "step": 27198
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8094447930806392e-07,
+      "loss": 0.267,
+      "step": 27199
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8073230942829646e-07,
+      "loss": 0.3103,
+      "step": 27200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8052026287908253e-07,
+      "loss": 0.2979,
+      "step": 27201
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8030833966308226e-07,
+      "loss": 0.2634,
+      "step": 27202
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8009653978296017e-07,
+      "loss": 0.3172,
+      "step": 27203
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7988486324137523e-07,
+      "loss": 0.2884,
+      "step": 27204
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7967331004098643e-07,
+      "loss": 0.3281,
+      "step": 27205
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7946188018445054e-07,
+      "loss": 0.3069,
+      "step": 27206
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7925057367442434e-07,
+      "loss": 0.3167,
+      "step": 27207
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7903939051356013e-07,
+      "loss": 0.298,
+      "step": 27208
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7882833070451134e-07,
+      "loss": 0.3127,
+      "step": 27209
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.786173942499303e-07,
+      "loss": 0.2926,
+      "step": 27210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.784065811524638e-07,
+      "loss": 0.3018,
+      "step": 27211
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7819589141475968e-07,
+      "loss": 0.3057,
+      "step": 27212
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7798532503946586e-07,
+      "loss": 0.311,
+      "step": 27213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7777488202922688e-07,
+      "loss": 0.3073,
+      "step": 27214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7756456238668506e-07,
+      "loss": 0.2985,
+      "step": 27215
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7735436611448277e-07,
+      "loss": 0.3551,
+      "step": 27216
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7714429321526005e-07,
+      "loss": 0.2664,
+      "step": 27217
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7693434369165374e-07,
+      "loss": 0.2762,
+      "step": 27218
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7672451754630283e-07,
+      "loss": 0.2917,
+      "step": 27219
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7651481478184297e-07,
+      "loss": 0.3009,
+      "step": 27220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.763052354009076e-07,
+      "loss": 0.2917,
+      "step": 27221
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7609577940612687e-07,
+      "loss": 0.2698,
+      "step": 27222
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7588644680013535e-07,
+      "loss": 0.3166,
+      "step": 27223
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7567723758555865e-07,
+      "loss": 0.3227,
+      "step": 27224
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7546815176502696e-07,
+      "loss": 0.3247,
+      "step": 27225
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7525918934116483e-07,
+      "loss": 0.2775,
+      "step": 27226
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7505035031659788e-07,
+      "loss": 0.2998,
+      "step": 27227
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7484163469394854e-07,
+      "loss": 0.2929,
+      "step": 27228
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7463304247583912e-07,
+      "loss": 0.3102,
+      "step": 27229
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7442457366488864e-07,
+      "loss": 0.2907,
+      "step": 27230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.742162282637161e-07,
+      "loss": 0.332,
+      "step": 27231
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7400800627493718e-07,
+      "loss": 0.3026,
+      "step": 27232
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7379990770116984e-07,
+      "loss": 0.3013,
+      "step": 27233
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7359193254502415e-07,
+      "loss": 0.321,
+      "step": 27234
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7338408080911473e-07,
+      "loss": 0.3213,
+      "step": 27235
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7317635249605168e-07,
+      "loss": 0.3153,
+      "step": 27236
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7296874760844296e-07,
+      "loss": 0.3474,
+      "step": 27237
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7276126614889866e-07,
+      "loss": 0.2838,
+      "step": 27238
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7255390812002227e-07,
+      "loss": 0.3341,
+      "step": 27239
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7234667352441947e-07,
+      "loss": 0.3212,
+      "step": 27240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7213956236469375e-07,
+      "loss": 0.2994,
+      "step": 27241
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7193257464344416e-07,
+      "loss": 0.2953,
+      "step": 27242
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7172571036327302e-07,
+      "loss": 0.3129,
+      "step": 27243
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7151896952677716e-07,
+      "loss": 0.3104,
+      "step": 27244
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7131235213655338e-07,
+      "loss": 0.3117,
+      "step": 27245
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7110585819519632e-07,
+      "loss": 0.3232,
+      "step": 27246
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.708994877053005e-07,
+      "loss": 0.3138,
+      "step": 27247
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7069324066945837e-07,
+      "loss": 0.2718,
+      "step": 27248
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.704871170902589e-07,
+      "loss": 0.315,
+      "step": 27249
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7028111697029116e-07,
+      "loss": 0.3073,
+      "step": 27250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7007524031214418e-07,
+      "loss": 0.2992,
+      "step": 27251
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6986948711840257e-07,
+      "loss": 0.3157,
+      "step": 27252
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6966385739165092e-07,
+      "loss": 0.2372,
+      "step": 27253
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6945835113447163e-07,
+      "loss": 0.2986,
+      "step": 27254
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6925296834944482e-07,
+      "loss": 0.2804,
+      "step": 27255
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.690477090391529e-07,
+      "loss": 0.3279,
+      "step": 27256
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.688425732061716e-07,
+      "loss": 0.3125,
+      "step": 27257
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.686375608530777e-07,
+      "loss": 0.3027,
+      "step": 27258
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6843267198244695e-07,
+      "loss": 0.2466,
+      "step": 27259
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6822790659685395e-07,
+      "loss": 0.3307,
+      "step": 27260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6802326469886666e-07,
+      "loss": 0.293,
+      "step": 27261
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6781874629105967e-07,
+      "loss": 0.26,
+      "step": 27262
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.676143513759998e-07,
+      "loss": 0.3196,
+      "step": 27263
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.674100799562539e-07,
+      "loss": 0.3277,
+      "step": 27264
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6720593203438772e-07,
+      "loss": 0.3078,
+      "step": 27265
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6700190761296697e-07,
+      "loss": 0.3129,
+      "step": 27266
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6679800669455072e-07,
+      "loss": 0.337,
+      "step": 27267
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6659422928170354e-07,
+      "loss": 0.275,
+      "step": 27268
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6639057537698455e-07,
+      "loss": 0.2695,
+      "step": 27269
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6618704498294947e-07,
+      "loss": 0.3289,
+      "step": 27270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.659836381021551e-07,
+      "loss": 0.3042,
+      "step": 27271
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6578035473715835e-07,
+      "loss": 0.3285,
+      "step": 27272
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.655771948905105e-07,
+      "loss": 0.2754,
+      "step": 27273
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6537415856476501e-07,
+      "loss": 0.307,
+      "step": 27274
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.651712457624699e-07,
+      "loss": 0.3433,
+      "step": 27275
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6496845648617532e-07,
+      "loss": 0.2958,
+      "step": 27276
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6476579073842703e-07,
+      "loss": 0.3152,
+      "step": 27277
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6456324852177075e-07,
+      "loss": 0.3061,
+      "step": 27278
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6436082983875223e-07,
+      "loss": 0.3087,
+      "step": 27279
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6415853469191169e-07,
+      "loss": 0.2824,
+      "step": 27280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6395636308379036e-07,
+      "loss": 0.2969,
+      "step": 27281
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6375431501692962e-07,
+      "loss": 0.3128,
+      "step": 27282
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6355239049386406e-07,
+      "loss": 0.3208,
+      "step": 27283
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.633505895171328e-07,
+      "loss": 0.2839,
+      "step": 27284
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.631489120892682e-07,
+      "loss": 0.2984,
+      "step": 27285
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6294735821280494e-07,
+      "loss": 0.3162,
+      "step": 27286
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.627459278902721e-07,
+      "loss": 0.2927,
+      "step": 27287
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6254462112420323e-07,
+      "loss": 0.2749,
+      "step": 27288
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6234343791712403e-07,
+      "loss": 0.2908,
+      "step": 27289
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6214237827156143e-07,
+      "loss": 0.3315,
+      "step": 27290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6194144219004227e-07,
+      "loss": 0.332,
+      "step": 27291
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6174062967509007e-07,
+      "loss": 0.3066,
+      "step": 27292
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.615399407292251e-07,
+      "loss": 0.3118,
+      "step": 27293
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6133937535497192e-07,
+      "loss": 0.3182,
+      "step": 27294
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6113893355484522e-07,
+      "loss": 0.2999,
+      "step": 27295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6093861533136523e-07,
+      "loss": 0.295,
+      "step": 27296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.607384206870477e-07,
+      "loss": 0.3076,
+      "step": 27297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6053834962440617e-07,
+      "loss": 0.2942,
+      "step": 27298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6033840214595309e-07,
+      "loss": 0.2952,
+      "step": 27299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.601385782542031e-07,
+      "loss": 0.3043,
+      "step": 27300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5993887795166197e-07,
+      "loss": 0.2906,
+      "step": 27301
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5973930124083882e-07,
+      "loss": 0.3129,
+      "step": 27302
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5953984812424273e-07,
+      "loss": 0.3215,
+      "step": 27303
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5934051860437727e-07,
+      "loss": 0.3199,
+      "step": 27304
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.59141312683746e-07,
+      "loss": 0.3089,
+      "step": 27305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5894223036485024e-07,
+      "loss": 0.3243,
+      "step": 27306
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5874327165019132e-07,
+      "loss": 0.3056,
+      "step": 27307
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5854443654226836e-07,
+      "loss": 0.3066,
+      "step": 27308
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5834572504357825e-07,
+      "loss": 0.301,
+      "step": 27309
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.581471371566179e-07,
+      "loss": 0.311,
+      "step": 27310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.579486728838797e-07,
+      "loss": 0.3176,
+      "step": 27311
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.577503322278573e-07,
+      "loss": 0.3042,
+      "step": 27312
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5755211519104196e-07,
+      "loss": 0.2804,
+      "step": 27313
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5735402177592397e-07,
+      "loss": 0.3083,
+      "step": 27314
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5715605198499128e-07,
+      "loss": 0.3236,
+      "step": 27315
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5695820582072863e-07,
+      "loss": 0.3374,
+      "step": 27316
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5676048328562289e-07,
+      "loss": 0.3014,
+      "step": 27317
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5656288438215538e-07,
+      "loss": 0.2917,
+      "step": 27318
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5636540911281085e-07,
+      "loss": 0.3195,
+      "step": 27319
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5616805748006725e-07,
+      "loss": 0.3326,
+      "step": 27320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5597082948640486e-07,
+      "loss": 0.2818,
+      "step": 27321
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5577372513430055e-07,
+      "loss": 0.3141,
+      "step": 27322
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5557674442622795e-07,
+      "loss": 0.301,
+      "step": 27323
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5537988736466393e-07,
+      "loss": 0.2958,
+      "step": 27324
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5518315395207982e-07,
+      "loss": 0.2981,
+      "step": 27325
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5498654419094706e-07,
+      "loss": 0.3339,
+      "step": 27326
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.547900580837336e-07,
+      "loss": 0.2882,
+      "step": 27327
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.545936956329097e-07,
+      "loss": 0.2806,
+      "step": 27328
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5439745684093898e-07,
+      "loss": 0.2999,
+      "step": 27329
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5420134171028833e-07,
+      "loss": 0.2991,
+      "step": 27330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5400535024342023e-07,
+      "loss": 0.3006,
+      "step": 27331
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5380948244279603e-07,
+      "loss": 0.3161,
+      "step": 27332
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.53613738310876e-07,
+      "loss": 0.2665,
+      "step": 27333
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.534181178501204e-07,
+      "loss": 0.3094,
+      "step": 27334
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.532226210629828e-07,
+      "loss": 0.3125,
+      "step": 27335
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5302724795192237e-07,
+      "loss": 0.3073,
+      "step": 27336
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5283199851939046e-07,
+      "loss": 0.319,
+      "step": 27337
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5263687276783955e-07,
+      "loss": 0.281,
+      "step": 27338
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5244187069972104e-07,
+      "loss": 0.3058,
+      "step": 27339
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5224699231748518e-07,
+      "loss": 0.3247,
+      "step": 27340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.520522376235778e-07,
+      "loss": 0.2959,
+      "step": 27341
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.518576066204458e-07,
+      "loss": 0.3089,
+      "step": 27342
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5166309931053503e-07,
+      "loss": 0.278,
+      "step": 27343
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5146871569628574e-07,
+      "loss": 0.2897,
+      "step": 27344
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5127445578014156e-07,
+      "loss": 0.2981,
+      "step": 27345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.510803195645416e-07,
+      "loss": 0.3241,
+      "step": 27346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5088630705192507e-07,
+      "loss": 0.3018,
+      "step": 27347
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.506924182447278e-07,
+      "loss": 0.3131,
+      "step": 27348
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5049865314538449e-07,
+      "loss": 0.3076,
+      "step": 27349
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5030501175633206e-07,
+      "loss": 0.2849,
+      "step": 27350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5011149407999748e-07,
+      "loss": 0.3026,
+      "step": 27351
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4991810011881548e-07,
+      "loss": 0.3182,
+      "step": 27352
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.497248298752141e-07,
+      "loss": 0.2971,
+      "step": 27353
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4953168335162027e-07,
+      "loss": 0.3332,
+      "step": 27354
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4933866055045987e-07,
+      "loss": 0.3005,
+      "step": 27355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.491457614741587e-07,
+      "loss": 0.3088,
+      "step": 27356
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.489529861251371e-07,
+      "loss": 0.2843,
+      "step": 27357
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4876033450581863e-07,
+      "loss": 0.3355,
+      "step": 27358
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4856780661862246e-07,
+      "loss": 0.2999,
+      "step": 27359
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.483754024659656e-07,
+      "loss": 0.3268,
+      "step": 27360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4818312205026496e-07,
+      "loss": 0.3203,
+      "step": 27361
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4799096537393754e-07,
+      "loss": 0.3255,
+      "step": 27362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4779893243939358e-07,
+      "loss": 0.3103,
+      "step": 27363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4760702324904787e-07,
+      "loss": 0.3014,
+      "step": 27364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4741523780530952e-07,
+      "loss": 0.3225,
+      "step": 27365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.472235761105878e-07,
+      "loss": 0.3099,
+      "step": 27366
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.470320381672885e-07,
+      "loss": 0.2967,
+      "step": 27367
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4684062397782084e-07,
+      "loss": 0.3328,
+      "step": 27368
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4664933354458622e-07,
+      "loss": 0.3146,
+      "step": 27369
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4645816686998716e-07,
+      "loss": 0.3024,
+      "step": 27370
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4626712395642616e-07,
+      "loss": 0.3383,
+      "step": 27371
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4607620480630026e-07,
+      "loss": 0.3087,
+      "step": 27372
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.458854094220097e-07,
+      "loss": 0.2909,
+      "step": 27373
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4569473780595146e-07,
+      "loss": 0.3581,
+      "step": 27374
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4550418996051808e-07,
+      "loss": 0.2833,
+      "step": 27375
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.453137658881043e-07,
+      "loss": 0.2717,
+      "step": 27376
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4512346559110159e-07,
+      "loss": 0.2836,
+      "step": 27377
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.449332890719002e-07,
+      "loss": 0.3154,
+      "step": 27378
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4474323633288935e-07,
+      "loss": 0.3013,
+      "step": 27379
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4455330737645489e-07,
+      "loss": 0.3102,
+      "step": 27380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4436350220498274e-07,
+      "loss": 0.2968,
+      "step": 27381
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4417382082085652e-07,
+      "loss": 0.3239,
+      "step": 27382
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4398426322645987e-07,
+      "loss": 0.3076,
+      "step": 27383
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.43794829424172e-07,
+      "loss": 0.2809,
+      "step": 27384
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4360551941637323e-07,
+      "loss": 0.3143,
+      "step": 27385
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4341633320544057e-07,
+      "loss": 0.3248,
+      "step": 27386
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.432272707937521e-07,
+      "loss": 0.2855,
+      "step": 27387
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4303833218368034e-07,
+      "loss": 0.3127,
+      "step": 27388
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4284951737760012e-07,
+      "loss": 0.298,
+      "step": 27389
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.426608263778828e-07,
+      "loss": 0.2982,
+      "step": 27390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4247225918689546e-07,
+      "loss": 0.2808,
+      "step": 27391
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4228381580700945e-07,
+      "loss": 0.2922,
+      "step": 27392
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.420954962405907e-07,
+      "loss": 0.3337,
+      "step": 27393
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.41907300490004e-07,
+      "loss": 0.3103,
+      "step": 27394
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.417192285576141e-07,
+      "loss": 0.2989,
+      "step": 27395
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4153128044578358e-07,
+      "loss": 0.3128,
+      "step": 27396
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4134345615687162e-07,
+      "loss": 0.3193,
+      "step": 27397
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4115575569323859e-07,
+      "loss": 0.2896,
+      "step": 27398
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4096817905724146e-07,
+      "loss": 0.3276,
+      "step": 27399
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4078072625123617e-07,
+      "loss": 0.2832,
+      "step": 27400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4059339727757638e-07,
+      "loss": 0.329,
+      "step": 27401
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4040619213861684e-07,
+      "loss": 0.3143,
+      "step": 27402
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4021911083670682e-07,
+      "loss": 0.2975,
+      "step": 27403
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4003215337419662e-07,
+      "loss": 0.3012,
+      "step": 27404
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3984531975343553e-07,
+      "loss": 0.3321,
+      "step": 27405
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3965860997676828e-07,
+      "loss": 0.3035,
+      "step": 27406
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3947202404654192e-07,
+      "loss": 0.2967,
+      "step": 27407
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3928556196510012e-07,
+      "loss": 0.2952,
+      "step": 27408
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3909922373478325e-07,
+      "loss": 0.3026,
+      "step": 27409
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3891300935793272e-07,
+      "loss": 0.319,
+      "step": 27410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3872691883688673e-07,
+      "loss": 0.2884,
+      "step": 27411
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3854095217398222e-07,
+      "loss": 0.2972,
+      "step": 27412
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3835510937155627e-07,
+      "loss": 0.2954,
+      "step": 27413
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3816939043194256e-07,
+      "loss": 0.3286,
+      "step": 27414
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3798379535747363e-07,
+      "loss": 0.3235,
+      "step": 27415
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.377983241504799e-07,
+      "loss": 0.3227,
+      "step": 27416
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.376129768132928e-07,
+      "loss": 0.3102,
+      "step": 27417
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.374277533482382e-07,
+      "loss": 0.2978,
+      "step": 27418
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3724265375764322e-07,
+      "loss": 0.3201,
+      "step": 27419
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3705767804383375e-07,
+      "loss": 0.2825,
+      "step": 27420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3687282620913234e-07,
+      "loss": 0.306,
+      "step": 27421
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3668809825585938e-07,
+      "loss": 0.2779,
+      "step": 27422
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3650349418633635e-07,
+      "loss": 0.2993,
+      "step": 27423
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.363190140028836e-07,
+      "loss": 0.2733,
+      "step": 27424
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3613465770781487e-07,
+      "loss": 0.2987,
+      "step": 27425
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3595042530344715e-07,
+      "loss": 0.318,
+      "step": 27426
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.357663167920953e-07,
+      "loss": 0.3095,
+      "step": 27427
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3558233217607074e-07,
+      "loss": 0.3396,
+      "step": 27428
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3539847145768392e-07,
+      "loss": 0.3192,
+      "step": 27429
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.352147346392463e-07,
+      "loss": 0.3016,
+      "step": 27430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3503112172306377e-07,
+      "loss": 0.3256,
+      "step": 27431
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.348476327114423e-07,
+      "loss": 0.331,
+      "step": 27432
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3466426760668672e-07,
+      "loss": 0.3189,
+      "step": 27433
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3448102641110073e-07,
+      "loss": 0.3004,
+      "step": 27434
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3429790912698582e-07,
+      "loss": 0.2909,
+      "step": 27435
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3411491575664237e-07,
+      "loss": 0.3073,
+      "step": 27436
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3393204630236632e-07,
+      "loss": 0.2895,
+      "step": 27437
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3374930076645697e-07,
+      "loss": 0.2665,
+      "step": 27438
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3356667915121025e-07,
+      "loss": 0.3152,
+      "step": 27439
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3338418145891763e-07,
+      "loss": 0.3205,
+      "step": 27440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3320180769187286e-07,
+      "loss": 0.3076,
+      "step": 27441
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.330195578523652e-07,
+      "loss": 0.2718,
+      "step": 27442
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3283743194268506e-07,
+      "loss": 0.2862,
+      "step": 27443
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3265542996511838e-07,
+      "loss": 0.296,
+      "step": 27444
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3247355192195331e-07,
+      "loss": 0.2704,
+      "step": 27445
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3229179781547254e-07,
+      "loss": 0.3259,
+      "step": 27446
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.321101676479597e-07,
+      "loss": 0.3055,
+      "step": 27447
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3192866142169636e-07,
+      "loss": 0.3375,
+      "step": 27448
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3174727913896068e-07,
+      "loss": 0.2813,
+      "step": 27449
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3156602080203196e-07,
+      "loss": 0.2998,
+      "step": 27450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3138488641318724e-07,
+      "loss": 0.2818,
+      "step": 27451
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.312038759747003e-07,
+      "loss": 0.2634,
+      "step": 27452
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3102298948884596e-07,
+      "loss": 0.3113,
+      "step": 27453
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3084222695789683e-07,
+      "loss": 0.3153,
+      "step": 27454
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3066158838412002e-07,
+      "loss": 0.304,
+      "step": 27455
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3048107376978703e-07,
+      "loss": 0.2797,
+      "step": 27456
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3030068311716492e-07,
+      "loss": 0.2707,
+      "step": 27457
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3012041642851858e-07,
+      "loss": 0.2975,
+      "step": 27458
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2994027370611173e-07,
+      "loss": 0.3144,
+      "step": 27459
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2976025495220923e-07,
+      "loss": 0.3204,
+      "step": 27460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2958036016907038e-07,
+      "loss": 0.3148,
+      "step": 27461
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.294005893589545e-07,
+      "loss": 0.3147,
+      "step": 27462
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2922094252412087e-07,
+      "loss": 0.3095,
+      "step": 27463
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2904141966682548e-07,
+      "loss": 0.2989,
+      "step": 27464
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2886202078932097e-07,
+      "loss": 0.2913,
+      "step": 27465
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2868274589386332e-07,
+      "loss": 0.2996,
+      "step": 27466
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2850359498270294e-07,
+      "loss": 0.3279,
+      "step": 27467
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2832456805809023e-07,
+      "loss": 0.2951,
+      "step": 27468
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2814566512227343e-07,
+      "loss": 0.297,
+      "step": 27469
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2796688617750185e-07,
+      "loss": 0.3064,
+      "step": 27470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2778823122601703e-07,
+      "loss": 0.3193,
+      "step": 27471
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2760970027006493e-07,
+      "loss": 0.2916,
+      "step": 27472
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.274312933118893e-07,
+      "loss": 0.3135,
+      "step": 27473
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.272530103537295e-07,
+      "loss": 0.2858,
+      "step": 27474
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2707485139782372e-07,
+      "loss": 0.3204,
+      "step": 27475
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2689681644641238e-07,
+      "loss": 0.3042,
+      "step": 27476
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2671890550172817e-07,
+      "loss": 0.3096,
+      "step": 27477
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2654111856600814e-07,
+      "loss": 0.2888,
+      "step": 27478
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.26363455641485e-07,
+      "loss": 0.2916,
+      "step": 27479
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2618591673038915e-07,
+      "loss": 0.3045,
+      "step": 27480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2600850183495107e-07,
+      "loss": 0.3218,
+      "step": 27481
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2583121095740002e-07,
+      "loss": 0.2344,
+      "step": 27482
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2565404409996095e-07,
+      "loss": 0.3288,
+      "step": 27483
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2547700126486208e-07,
+      "loss": 0.2699,
+      "step": 27484
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.253000824543238e-07,
+      "loss": 0.3299,
+      "step": 27485
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.251232876705688e-07,
+      "loss": 0.3155,
+      "step": 27486
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2494661691581978e-07,
+      "loss": 0.3164,
+      "step": 27487
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.247700701922938e-07,
+      "loss": 0.3266,
+      "step": 27488
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2459364750220803e-07,
+      "loss": 0.3077,
+      "step": 27489
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2441734884777957e-07,
+      "loss": 0.2745,
+      "step": 27490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.242411742312233e-07,
+      "loss": 0.3231,
+      "step": 27491
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2406512365474965e-07,
+      "loss": 0.2858,
+      "step": 27492
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2388919712057136e-07,
+      "loss": 0.3197,
+      "step": 27493
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2371339463089882e-07,
+      "loss": 0.2962,
+      "step": 27494
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.235377161879392e-07,
+      "loss": 0.3034,
+      "step": 27495
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2336216179389737e-07,
+      "loss": 0.3367,
+      "step": 27496
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.231867314509805e-07,
+      "loss": 0.3551,
+      "step": 27497
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.230114251613923e-07,
+      "loss": 0.3404,
+      "step": 27498
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2283624292733332e-07,
+      "loss": 0.3048,
+      "step": 27499
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2266118475100397e-07,
+      "loss": 0.3047,
+      "step": 27500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2248625063460474e-07,
+      "loss": 0.3426,
+      "step": 27501
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.223114405803294e-07,
+      "loss": 0.2735,
+      "step": 27502
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.221367545903762e-07,
+      "loss": 0.2844,
+      "step": 27503
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2196219266694008e-07,
+      "loss": 0.325,
+      "step": 27504
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.217877548122115e-07,
+      "loss": 0.2967,
+      "step": 27505
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.216134410283809e-07,
+      "loss": 0.2869,
+      "step": 27506
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2143925131763877e-07,
+      "loss": 0.3255,
+      "step": 27507
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.212651856821734e-07,
+      "loss": 0.2797,
+      "step": 27508
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2109124412417074e-07,
+      "loss": 0.3056,
+      "step": 27509
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2091742664581575e-07,
+      "loss": 0.308,
+      "step": 27510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2074373324929e-07,
+      "loss": 0.3149,
+      "step": 27511
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2057016393677623e-07,
+      "loss": 0.3186,
+      "step": 27512
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.20396718710456e-07,
+      "loss": 0.306,
+      "step": 27513
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2022339757250424e-07,
+      "loss": 0.345,
+      "step": 27514
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.200502005251014e-07,
+      "loss": 0.2929,
+      "step": 27515
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1987712757042136e-07,
+      "loss": 0.3119,
+      "step": 27516
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.197041787106379e-07,
+      "loss": 0.3153,
+      "step": 27517
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.195313539479226e-07,
+      "loss": 0.3276,
+      "step": 27518
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1935865328444818e-07,
+      "loss": 0.2929,
+      "step": 27519
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.191860767223807e-07,
+      "loss": 0.2884,
+      "step": 27520
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1901362426389062e-07,
+      "loss": 0.3062,
+      "step": 27521
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1884129591114291e-07,
+      "loss": 0.3052,
+      "step": 27522
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1866909166630136e-07,
+      "loss": 0.2868,
+      "step": 27523
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.184970115315287e-07,
+      "loss": 0.313,
+      "step": 27524
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1832505550898877e-07,
+      "loss": 0.2751,
+      "step": 27525
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1815322360083869e-07,
+      "loss": 0.3018,
+      "step": 27526
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1798151580923789e-07,
+      "loss": 0.309,
+      "step": 27527
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1780993213634239e-07,
+      "loss": 0.3026,
+      "step": 27528
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1763847258430828e-07,
+      "loss": 0.2777,
+      "step": 27529
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1746713715528713e-07,
+      "loss": 0.3118,
+      "step": 27530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1729592585143279e-07,
+      "loss": 0.2875,
+      "step": 27531
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1712483867489577e-07,
+      "loss": 0.3022,
+      "step": 27532
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1695387562782323e-07,
+      "loss": 0.3247,
+      "step": 27533
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1678303671236346e-07,
+      "loss": 0.2909,
+      "step": 27534
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1661232193066363e-07,
+      "loss": 0.3101,
+      "step": 27535
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1644173128486535e-07,
+      "loss": 0.3164,
+      "step": 27536
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.162712647771136e-07,
+      "loss": 0.3216,
+      "step": 27537
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1610092240954773e-07,
+      "loss": 0.3161,
+      "step": 27538
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1593070418430719e-07,
+      "loss": 0.2689,
+      "step": 27539
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1576061010353134e-07,
+      "loss": 0.3049,
+      "step": 27540
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1559064016935628e-07,
+      "loss": 0.2746,
+      "step": 27541
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1542079438391474e-07,
+      "loss": 0.3054,
+      "step": 27542
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1525107274934277e-07,
+      "loss": 0.2698,
+      "step": 27543
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1508147526777091e-07,
+      "loss": 0.3222,
+      "step": 27544
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1491200194132857e-07,
+      "loss": 0.2675,
+      "step": 27545
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1474265277214514e-07,
+      "loss": 0.3057,
+      "step": 27546
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1457342776234893e-07,
+      "loss": 0.2798,
+      "step": 27547
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1440432691406377e-07,
+      "loss": 0.2935,
+      "step": 27548
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1423535022941357e-07,
+      "loss": 0.3107,
+      "step": 27549
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1406649771052103e-07,
+      "loss": 0.2867,
+      "step": 27550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.138977693595067e-07,
+      "loss": 0.3037,
+      "step": 27551
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1372916517848997e-07,
+      "loss": 0.2799,
+      "step": 27552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1356068516958918e-07,
+      "loss": 0.2903,
+      "step": 27553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1339232933491928e-07,
+      "loss": 0.3433,
+      "step": 27554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1322409767659526e-07,
+      "loss": 0.3216,
+      "step": 27555
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1305599019673208e-07,
+      "loss": 0.2761,
+      "step": 27556
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1288800689743695e-07,
+      "loss": 0.3118,
+      "step": 27557
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1272014778082374e-07,
+      "loss": 0.3136,
+      "step": 27558
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1255241284899965e-07,
+      "loss": 0.3022,
+      "step": 27559
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1238480210406965e-07,
+      "loss": 0.2847,
+      "step": 27560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1221731554814097e-07,
+      "loss": 0.2555,
+      "step": 27561
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1204995318331636e-07,
+      "loss": 0.2978,
+      "step": 27562
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1188271501169857e-07,
+      "loss": 0.2994,
+      "step": 27563
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1171560103538704e-07,
+      "loss": 0.3305,
+      "step": 27564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1154861125648231e-07,
+      "loss": 0.2981,
+      "step": 27565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1138174567707937e-07,
+      "loss": 0.3551,
+      "step": 27566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1121500429927656e-07,
+      "loss": 0.303,
+      "step": 27567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1104838712516663e-07,
+      "loss": 0.2723,
+      "step": 27568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1088189415684459e-07,
+      "loss": 0.3481,
+      "step": 27569
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1071552539639762e-07,
+      "loss": 0.3345,
+      "step": 27570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1054928084591743e-07,
+      "loss": 0.2893,
+      "step": 27571
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1038316050749343e-07,
+      "loss": 0.2796,
+      "step": 27572
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1021716438321062e-07,
+      "loss": 0.3092,
+      "step": 27573
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.100512924751529e-07,
+      "loss": 0.2914,
+      "step": 27574
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0988554478540636e-07,
+      "loss": 0.3069,
+      "step": 27575
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0971992131605048e-07,
+      "loss": 0.2933,
+      "step": 27576
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0955442206916578e-07,
+      "loss": 0.3191,
+      "step": 27577
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0938904704683283e-07,
+      "loss": 0.328,
+      "step": 27578
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0922379625112777e-07,
+      "loss": 0.3025,
+      "step": 27579
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0905866968412449e-07,
+      "loss": 0.3136,
+      "step": 27580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0889366734789908e-07,
+      "loss": 0.3088,
+      "step": 27581
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0872878924452213e-07,
+      "loss": 0.2778,
+      "step": 27582
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.085640353760653e-07,
+      "loss": 0.3367,
+      "step": 27583
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0839940574459918e-07,
+      "loss": 0.3393,
+      "step": 27584
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0823490035218986e-07,
+      "loss": 0.3369,
+      "step": 27585
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0807051920090461e-07,
+      "loss": 0.3009,
+      "step": 27586
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0790626229280842e-07,
+      "loss": 0.315,
+      "step": 27587
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0774212962996189e-07,
+      "loss": 0.3029,
+      "step": 27588
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0757812121443e-07,
+      "loss": 0.3063,
+      "step": 27589
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0741423704827004e-07,
+      "loss": 0.317,
+      "step": 27590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0725047713354142e-07,
+      "loss": 0.3048,
+      "step": 27591
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0708684147230031e-07,
+      "loss": 0.3056,
+      "step": 27592
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0692333006660394e-07,
+      "loss": 0.2697,
+      "step": 27593
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0675994291850289e-07,
+      "loss": 0.3031,
+      "step": 27594
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0659668003005219e-07,
+      "loss": 0.2904,
+      "step": 27595
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.064335414033013e-07,
+      "loss": 0.3042,
+      "step": 27596
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0627052704029862e-07,
+      "loss": 0.3292,
+      "step": 27597
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0610763694309246e-07,
+      "loss": 0.3074,
+      "step": 27598
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0594487111372898e-07,
+      "loss": 0.2812,
+      "step": 27599
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.057822295542521e-07,
+      "loss": 0.2564,
+      "step": 27600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0561971226670353e-07,
+      "loss": 0.299,
+      "step": 27601
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0545731925312608e-07,
+      "loss": 0.3392,
+      "step": 27602
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0529505051555922e-07,
+      "loss": 0.3037,
+      "step": 27603
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.051329060560391e-07,
+      "loss": 0.2799,
+      "step": 27604
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.049708858766052e-07,
+      "loss": 0.3077,
+      "step": 27605
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0480898997929146e-07,
+      "loss": 0.2814,
+      "step": 27606
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0464721836613067e-07,
+      "loss": 0.2977,
+      "step": 27607
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0448557103915458e-07,
+      "loss": 0.3181,
+      "step": 27608
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0432404800039375e-07,
+      "loss": 0.2968,
+      "step": 27609
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.041626492518788e-07,
+      "loss": 0.285,
+      "step": 27610
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0400137479563366e-07,
+      "loss": 0.3598,
+      "step": 27611
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.038402246336856e-07,
+      "loss": 0.35,
+      "step": 27612
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0367919876805855e-07,
+      "loss": 0.3011,
+      "step": 27613
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0351829720077534e-07,
+      "loss": 0.3154,
+      "step": 27614
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0335751993385657e-07,
+      "loss": 0.2914,
+      "step": 27615
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0319686696932174e-07,
+      "loss": 0.281,
+      "step": 27616
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0303633830918813e-07,
+      "loss": 0.302,
+      "step": 27617
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0287593395547301e-07,
+      "loss": 0.2952,
+      "step": 27618
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0271565391018922e-07,
+      "loss": 0.2943,
+      "step": 27619
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0255549817535293e-07,
+      "loss": 0.2865,
+      "step": 27620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0239546675297251e-07,
+      "loss": 0.3424,
+      "step": 27621
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0223555964506082e-07,
+      "loss": 0.2889,
+      "step": 27622
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.020757768536229e-07,
+      "loss": 0.3144,
+      "step": 27623
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.019161183806694e-07,
+      "loss": 0.3177,
+      "step": 27624
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0175658422820201e-07,
+      "loss": 0.3123,
+      "step": 27625
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0159717439822692e-07,
+      "loss": 0.303,
+      "step": 27626
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0143788889274697e-07,
+      "loss": 0.3239,
+      "step": 27627
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0127872771375947e-07,
+      "loss": 0.3197,
+      "step": 27628
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0111969086326723e-07,
+      "loss": 0.3155,
+      "step": 27629
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0096077834326534e-07,
+      "loss": 0.2597,
+      "step": 27630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0080199015574998e-07,
+      "loss": 0.3303,
+      "step": 27631
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0064332630271733e-07,
+      "loss": 0.3183,
+      "step": 27632
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0048478678615803e-07,
+      "loss": 0.311,
+      "step": 27633
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.003263716080649e-07,
+      "loss": 0.332,
+      "step": 27634
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0016808077042639e-07,
+      "loss": 0.3182,
+      "step": 27635
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0000991427523087e-07,
+      "loss": 0.3003,
+      "step": 27636
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.985187212446568e-08,
+      "loss": 0.2901,
+      "step": 27637
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.969395432011586e-08,
+      "loss": 0.2844,
+      "step": 27638
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.953616086416429e-08,
+      "loss": 0.3015,
+      "step": 27639
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.937849175859271e-08,
+      "loss": 0.3363,
+      "step": 27640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.922094700538176e-08,
+      "loss": 0.3009,
+      "step": 27641
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.906352660651097e-08,
+      "loss": 0.2844,
+      "step": 27642
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.890623056395653e-08,
+      "loss": 0.317,
+      "step": 27643
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.874905887969355e-08,
+      "loss": 0.2963,
+      "step": 27644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.859201155569709e-08,
+      "loss": 0.3137,
+      "step": 27645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.843508859394001e-08,
+      "loss": 0.2793,
+      "step": 27646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.8278289996393e-08,
+      "loss": 0.3409,
+      "step": 27647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.812161576502444e-08,
+      "loss": 0.3185,
+      "step": 27648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.796506590180389e-08,
+      "loss": 0.3069,
+      "step": 27649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.78086404086953e-08,
+      "loss": 0.2887,
+      "step": 27650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.765233928766493e-08,
+      "loss": 0.3388,
+      "step": 27651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.749616254067784e-08,
+      "loss": 0.3173,
+      "step": 27652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.734011016969248e-08,
+      "loss": 0.3046,
+      "step": 27653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.718418217666837e-08,
+      "loss": 0.3,
+      "step": 27654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.702837856356728e-08,
+      "loss": 0.2823,
+      "step": 27655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.687269933234434e-08,
+      "loss": 0.3025,
+      "step": 27656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.67171444849535e-08,
+      "loss": 0.2933,
+      "step": 27657
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.656171402335213e-08,
+      "loss": 0.3525,
+      "step": 27658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.640640794948864e-08,
+      "loss": 0.3176,
+      "step": 27659
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.625122626531591e-08,
+      "loss": 0.3154,
+      "step": 27660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.60961689727824e-08,
+      "loss": 0.2998,
+      "step": 27661
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.594123607383543e-08,
+      "loss": 0.3264,
+      "step": 27662
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.578642757042123e-08,
+      "loss": 0.3182,
+      "step": 27663
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.563174346448489e-08,
+      "loss": 0.2956,
+      "step": 27664
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.547718375796822e-08,
+      "loss": 0.2877,
+      "step": 27665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.532274845281186e-08,
+      "loss": 0.2808,
+      "step": 27666
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.516843755095873e-08,
+      "loss": 0.3449,
+      "step": 27667
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.50142510543428e-08,
+      "loss": 0.311,
+      "step": 27668
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.486018896490257e-08,
+      "loss": 0.2722,
+      "step": 27669
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.470625128457423e-08,
+      "loss": 0.2994,
+      "step": 27670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.455243801528957e-08,
+      "loss": 0.2669,
+      "step": 27671
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.43987491589804e-08,
+      "loss": 0.2878,
+      "step": 27672
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.424518471757959e-08,
+      "loss": 0.3004,
+      "step": 27673
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.40917446930123e-08,
+      "loss": 0.2885,
+      "step": 27674
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.393842908720808e-08,
+      "loss": 0.322,
+      "step": 27675
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.378523790209315e-08,
+      "loss": 0.3083,
+      "step": 27676
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.363217113958822e-08,
+      "loss": 0.2879,
+      "step": 27677
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.347922880161952e-08,
+      "loss": 0.331,
+      "step": 27678
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.332641089010553e-08,
+      "loss": 0.312,
+      "step": 27679
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.317371740696801e-08,
+      "loss": 0.2863,
+      "step": 27680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.302114835412213e-08,
+      "loss": 0.3036,
+      "step": 27681
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.286870373348522e-08,
+      "loss": 0.3018,
+      "step": 27682
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.271638354697354e-08,
+      "loss": 0.3252,
+      "step": 27683
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.256418779649889e-08,
+      "loss": 0.274,
+      "step": 27684
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.241211648397197e-08,
+      "loss": 0.2345,
+      "step": 27685
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.226016961130458e-08,
+      "loss": 0.2958,
+      "step": 27686
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.210834718040406e-08,
+      "loss": 0.32,
+      "step": 27687
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.195664919317782e-08,
+      "loss": 0.3192,
+      "step": 27688
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.180507565153097e-08,
+      "loss": 0.2941,
+      "step": 27689
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.165362655736642e-08,
+      "loss": 0.2896,
+      "step": 27690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.150230191258824e-08,
+      "loss": 0.3228,
+      "step": 27691
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.1351101719096e-08,
+      "loss": 0.3124,
+      "step": 27692
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.120002597878819e-08,
+      "loss": 0.262,
+      "step": 27693
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.104907469356438e-08,
+      "loss": 0.2983,
+      "step": 27694
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.089824786531754e-08,
+      "loss": 0.3368,
+      "step": 27695
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.0747545495945e-08,
+      "loss": 0.28,
+      "step": 27696
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.059696758733749e-08,
+      "loss": 0.3107,
+      "step": 27697
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.044651414138683e-08,
+      "loss": 0.3084,
+      "step": 27698
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.029618515998373e-08,
+      "loss": 0.3044,
+      "step": 27699
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.014598064501445e-08,
+      "loss": 0.2924,
+      "step": 27700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.999590059836749e-08,
+      "loss": 0.3164,
+      "step": 27701
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.98459450219269e-08,
+      "loss": 0.3073,
+      "step": 27702
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.96961139175756e-08,
+      "loss": 0.3,
+      "step": 27703
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.954640728719766e-08,
+      "loss": 0.3227,
+      "step": 27704
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.939682513267155e-08,
+      "loss": 0.3416,
+      "step": 27705
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.92473674558758e-08,
+      "loss": 0.3139,
+      "step": 27706
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.909803425868779e-08,
+      "loss": 0.3195,
+      "step": 27707
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.894882554298379e-08,
+      "loss": 0.2753,
+      "step": 27708
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.879974131063785e-08,
+      "loss": 0.3119,
+      "step": 27709
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.865078156352291e-08,
+      "loss": 0.308,
+      "step": 27710
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.85019463035075e-08,
+      "loss": 0.2947,
+      "step": 27711
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.835323553246455e-08,
+      "loss": 0.2882,
+      "step": 27712
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.820464925225814e-08,
+      "loss": 0.3038,
+      "step": 27713
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.805618746475897e-08,
+      "loss": 0.2701,
+      "step": 27714
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.790785017182779e-08,
+      "loss": 0.3,
+      "step": 27715
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.775963737532867e-08,
+      "loss": 0.3255,
+      "step": 27716
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.761154907712454e-08,
+      "loss": 0.2804,
+      "step": 27717
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.746358527907394e-08,
+      "loss": 0.2998,
+      "step": 27718
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.731574598303649e-08,
+      "loss": 0.3079,
+      "step": 27719
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.716803119086847e-08,
+      "loss": 0.3117,
+      "step": 27720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.702044090442396e-08,
+      "loss": 0.3056,
+      "step": 27721
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.687297512555926e-08,
+      "loss": 0.3264,
+      "step": 27722
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.672563385612509e-08,
+      "loss": 0.3204,
+      "step": 27723
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.65784170979711e-08,
+      "loss": 0.3071,
+      "step": 27724
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.643132485294803e-08,
+      "loss": 0.3384,
+      "step": 27725
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.628435712290328e-08,
+      "loss": 0.2959,
+      "step": 27726
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.613751390968205e-08,
+      "loss": 0.3144,
+      "step": 27727
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.599079521512843e-08,
+      "loss": 0.2768,
+      "step": 27728
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.584420104108537e-08,
+      "loss": 0.3498,
+      "step": 27729
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.569773138939474e-08,
+      "loss": 0.295,
+      "step": 27730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.2605,
+      "step": 27731
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.540516566042711e-08,
+      "loss": 0.309,
+      "step": 27732
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.525906958682495e-08,
+      "loss": 0.3165,
+      "step": 27733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.511309804292267e-08,
+      "loss": 0.2917,
+      "step": 27734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.496725103055659e-08,
+      "loss": 0.3257,
+      "step": 27735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.482152855155745e-08,
+      "loss": 0.3445,
+      "step": 27736
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.467593060775492e-08,
+      "loss": 0.278,
+      "step": 27737
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.453045720097753e-08,
+      "loss": 0.3081,
+      "step": 27738
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.438510833305269e-08,
+      "loss": 0.301,
+      "step": 27739
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.423988400580674e-08,
+      "loss": 0.312,
+      "step": 27740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.409478422106376e-08,
+      "loss": 0.3397,
+      "step": 27741
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.394980898064453e-08,
+      "loss": 0.3332,
+      "step": 27742
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.380495828637092e-08,
+      "loss": 0.349,
+      "step": 27743
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.366023214006369e-08,
+      "loss": 0.3192,
+      "step": 27744
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.351563054353806e-08,
+      "loss": 0.34,
+      "step": 27745
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.337115349861257e-08,
+      "loss": 0.2838,
+      "step": 27746
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.322680100710023e-08,
+      "loss": 0.2576,
+      "step": 27747
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.308257307081513e-08,
+      "loss": 0.3251,
+      "step": 27748
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.293846969156694e-08,
+      "loss": 0.3359,
+      "step": 27749
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.279449087116865e-08,
+      "loss": 0.3079,
+      "step": 27750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.265063661142658e-08,
+      "loss": 0.3297,
+      "step": 27751
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.250690691414709e-08,
+      "loss": 0.3247,
+      "step": 27752
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.23633017811365e-08,
+      "loss": 0.3348,
+      "step": 27753
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.221982121419891e-08,
+      "loss": 0.3026,
+      "step": 27754
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.207646521513512e-08,
+      "loss": 0.297,
+      "step": 27755
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.193323378574702e-08,
+      "loss": 0.273,
+      "step": 27756
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.179012692783317e-08,
+      "loss": 0.3026,
+      "step": 27757
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.164714464319101e-08,
+      "loss": 0.329,
+      "step": 27758
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.150428693361467e-08,
+      "loss": 0.3238,
+      "step": 27759
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.13615538009016e-08,
+      "loss": 0.3321,
+      "step": 27760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.121894524684149e-08,
+      "loss": 0.2955,
+      "step": 27761
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.107646127322843e-08,
+      "loss": 0.3165,
+      "step": 27762
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.093410188184992e-08,
+      "loss": 0.3186,
+      "step": 27763
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.079186707449449e-08,
+      "loss": 0.2984,
+      "step": 27764
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.064975685294852e-08,
+      "loss": 0.3314,
+      "step": 27765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.050777121899722e-08,
+      "loss": 0.3068,
+      "step": 27766
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036591017442363e-08,
+      "loss": 0.2809,
+      "step": 27767
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.022417372101077e-08,
+      "loss": 0.3004,
+      "step": 27768
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.00825618605372e-08,
+      "loss": 0.3008,
+      "step": 27769
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.994107459478262e-08,
+      "loss": 0.3103,
+      "step": 27770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.979971192552338e-08,
+      "loss": 0.2753,
+      "step": 27771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.965847385453584e-08,
+      "loss": 0.3243,
+      "step": 27772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.951736038359193e-08,
+      "loss": 0.309,
+      "step": 27773
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.937637151446687e-08,
+      "loss": 0.2905,
+      "step": 27774
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.923550724893036e-08,
+      "loss": 0.3032,
+      "step": 27775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.909476758875101e-08,
+      "loss": 0.2949,
+      "step": 27776
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.895415253569628e-08,
+      "loss": 0.2963,
+      "step": 27777
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.881366209153473e-08,
+      "loss": 0.3158,
+      "step": 27778
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.867329625802833e-08,
+      "loss": 0.304,
+      "step": 27779
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.853305503694119e-08,
+      "loss": 0.2548,
+      "step": 27780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.839293843003415e-08,
+      "loss": 0.3398,
+      "step": 27781
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.8252946439068e-08,
+      "loss": 0.3033,
+      "step": 27782
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.811307906580023e-08,
+      "loss": 0.2924,
+      "step": 27783
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.797333631198834e-08,
+      "loss": 0.3105,
+      "step": 27784
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.783371817938645e-08,
+      "loss": 0.2951,
+      "step": 27785
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.769422466974874e-08,
+      "loss": 0.3499,
+      "step": 27786
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.755485578482714e-08,
+      "loss": 0.3038,
+      "step": 27787
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.741561152637245e-08,
+      "loss": 0.3077,
+      "step": 27788
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.727649189613439e-08,
+      "loss": 0.3184,
+      "step": 27789
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.713749689585825e-08,
+      "loss": 0.3394,
+      "step": 27790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.69986265272915e-08,
+      "loss": 0.3116,
+      "step": 27791
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.68598807921761e-08,
+      "loss": 0.2995,
+      "step": 27792
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.67212596922584e-08,
+      "loss": 0.2685,
+      "step": 27793
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.658276322927704e-08,
+      "loss": 0.2945,
+      "step": 27794
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.644439140497062e-08,
+      "loss": 0.3295,
+      "step": 27795
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.630614422107884e-08,
+      "loss": 0.2984,
+      "step": 27796
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.616802167933924e-08,
+      "loss": 0.3477,
+      "step": 27797
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.603002378148372e-08,
+      "loss": 0.2967,
+      "step": 27798
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.589215052924759e-08,
+      "loss": 0.3017,
+      "step": 27799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.575440192436167e-08,
+      "loss": 0.32,
+      "step": 27800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.561677796855571e-08,
+      "loss": 0.2981,
+      "step": 27801
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.547927866355942e-08,
+      "loss": 0.2704,
+      "step": 27802
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.53419040110992e-08,
+      "loss": 0.2811,
+      "step": 27803
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.520465401290033e-08,
+      "loss": 0.3052,
+      "step": 27804
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.5067528670687e-08,
+      "loss": 0.3139,
+      "step": 27805
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.493052798618117e-08,
+      "loss": 0.277,
+      "step": 27806
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.479365196110366e-08,
+      "loss": 0.3295,
+      "step": 27807
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.465690059717423e-08,
+      "loss": 0.3309,
+      "step": 27808
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.452027389610927e-08,
+      "loss": 0.2701,
+      "step": 27809
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.43837718596263e-08,
+      "loss": 0.3135,
+      "step": 27810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.42473944894384e-08,
+      "loss": 0.3275,
+      "step": 27811
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.411114178726087e-08,
+      "loss": 0.3365,
+      "step": 27812
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.39750137548012e-08,
+      "loss": 0.2924,
+      "step": 27813
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.383901039377139e-08,
+      "loss": 0.3169,
+      "step": 27814
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.370313170588117e-08,
+      "loss": 0.2666,
+      "step": 27815
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.356737769283362e-08,
+      "loss": 0.306,
+      "step": 27816
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.343174835633737e-08,
+      "loss": 0.3001,
+      "step": 27817
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.32962436980933e-08,
+      "loss": 0.3187,
+      "step": 27818
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.316086371980446e-08,
+      "loss": 0.294,
+      "step": 27819
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.302560842317175e-08,
+      "loss": 0.3187,
+      "step": 27820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.289047780989267e-08,
+      "loss": 0.3208,
+      "step": 27821
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.275547188166587e-08,
+      "loss": 0.3237,
+      "step": 27822
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.262059064018556e-08,
+      "loss": 0.3194,
+      "step": 27823
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.248583408714705e-08,
+      "loss": 0.2808,
+      "step": 27824
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.235120222424341e-08,
+      "loss": 0.2859,
+      "step": 27825
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.22166950531633e-08,
+      "loss": 0.2996,
+      "step": 27826
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.208231257559762e-08,
+      "loss": 0.3304,
+      "step": 27827
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.194805479323496e-08,
+      "loss": 0.3327,
+      "step": 27828
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.181392170775958e-08,
+      "loss": 0.3259,
+      "step": 27829
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.167991332085789e-08,
+      "loss": 0.2976,
+      "step": 27830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.154602963421298e-08,
+      "loss": 0.3131,
+      "step": 27831
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.141227064950573e-08,
+      "loss": 0.2905,
+      "step": 27832
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.127863636841592e-08,
+      "loss": 0.3,
+      "step": 27833
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.114512679262331e-08,
+      "loss": 0.3166,
+      "step": 27834
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.101174192380434e-08,
+      "loss": 0.3167,
+      "step": 27835
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.08784817636321e-08,
+      "loss": 0.3168,
+      "step": 27836
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.074534631378416e-08,
+      "loss": 0.32,
+      "step": 27837
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.061233557593028e-08,
+      "loss": 0.2972,
+      "step": 27838
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.047944955174025e-08,
+      "loss": 0.2942,
+      "step": 27839
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.034668824288604e-08,
+      "loss": 0.2961,
+      "step": 27840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.021405165103301e-08,
+      "loss": 0.3079,
+      "step": 27841
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.00815397778476e-08,
+      "loss": 0.2892,
+      "step": 27842
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.994915262499513e-08,
+      "loss": 0.3318,
+      "step": 27843
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.98168901941365e-08,
+      "loss": 0.2886,
+      "step": 27844
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.968475248693374e-08,
+      "loss": 0.3308,
+      "step": 27845
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.955273950504659e-08,
+      "loss": 0.2752,
+      "step": 27846
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.942085125013376e-08,
+      "loss": 0.3182,
+      "step": 27847
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.928908772385056e-08,
+      "loss": 0.3162,
+      "step": 27848
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.915744892785348e-08,
+      "loss": 0.3386,
+      "step": 27849
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.902593486379338e-08,
+      "loss": 0.2871,
+      "step": 27850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.889454553332453e-08,
+      "loss": 0.2803,
+      "step": 27851
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.87632809380978e-08,
+      "loss": 0.3058,
+      "step": 27852
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.863214107975969e-08,
+      "loss": 0.3252,
+      "step": 27853
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.850112595995773e-08,
+      "loss": 0.3139,
+      "step": 27854
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.837023558033728e-08,
+      "loss": 0.3213,
+      "step": 27855
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.823946994254371e-08,
+      "loss": 0.3506,
+      "step": 27856
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.810882904821791e-08,
+      "loss": 0.3151,
+      "step": 27857
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.797831289900303e-08,
+      "loss": 0.3209,
+      "step": 27858
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.784792149653552e-08,
+      "loss": 0.2963,
+      "step": 27859
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.771765484245407e-08,
+      "loss": 0.2651,
+      "step": 27860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.758751293839516e-08,
+      "loss": 0.2966,
+      "step": 27861
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.745749578599414e-08,
+      "loss": 0.2693,
+      "step": 27862
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.732760338688416e-08,
+      "loss": 0.3067,
+      "step": 27863
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.71978357426939e-08,
+      "loss": 0.3309,
+      "step": 27864
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.70681928550565e-08,
+      "loss": 0.3135,
+      "step": 27865
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.693867472559845e-08,
+      "loss": 0.3174,
+      "step": 27866
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.680928135594622e-08,
+      "loss": 0.3351,
+      "step": 27867
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.668001274772629e-08,
+      "loss": 0.2936,
+      "step": 27868
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.655086890256179e-08,
+      "loss": 0.3117,
+      "step": 27869
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.642184982207478e-08,
+      "loss": 0.3057,
+      "step": 27870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.629295550788506e-08,
+      "loss": 0.3257,
+      "step": 27871
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.616418596161245e-08,
+      "loss": 0.3063,
+      "step": 27872
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.603554118487454e-08,
+      "loss": 0.3362,
+      "step": 27873
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.590702117928672e-08,
+      "loss": 0.3129,
+      "step": 27874
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.577862594646323e-08,
+      "loss": 0.2571,
+      "step": 27875
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.565035548801501e-08,
+      "loss": 0.3369,
+      "step": 27876
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.552220980555635e-08,
+      "loss": 0.3107,
+      "step": 27877
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.539418890069482e-08,
+      "loss": 0.3174,
+      "step": 27878
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.526629277503804e-08,
+      "loss": 0.3161,
+      "step": 27879
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.513852143019362e-08,
+      "loss": 0.3194,
+      "step": 27880
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.501087486776469e-08,
+      "loss": 0.2967,
+      "step": 27881
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.488335308935666e-08,
+      "loss": 0.3012,
+      "step": 27882
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.475595609657048e-08,
+      "loss": 0.3008,
+      "step": 27883
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.462868389100485e-08,
+      "loss": 0.3099,
+      "step": 27884
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.450153647425961e-08,
+      "loss": 0.3266,
+      "step": 27885
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.437451384793236e-08,
+      "loss": 0.306,
+      "step": 27886
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.42476160136174e-08,
+      "loss": 0.3381,
+      "step": 27887
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.412084297290788e-08,
+      "loss": 0.3253,
+      "step": 27888
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.399419472739809e-08,
+      "loss": 0.3093,
+      "step": 27889
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.386767127867676e-08,
+      "loss": 0.3047,
+      "step": 27890
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.374127262833485e-08,
+      "loss": 0.295,
+      "step": 27891
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.361499877795885e-08,
+      "loss": 0.3186,
+      "step": 27892
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.348884972913416e-08,
+      "loss": 0.2806,
+      "step": 27893
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.33628254834473e-08,
+      "loss": 0.3218,
+      "step": 27894
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.323692604247921e-08,
+      "loss": 0.3359,
+      "step": 27895
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.311115140781199e-08,
+      "loss": 0.2929,
+      "step": 27896
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.298550158102545e-08,
+      "loss": 0.3141,
+      "step": 27897
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.285997656369725e-08,
+      "loss": 0.3073,
+      "step": 27898
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.273457635740388e-08,
+      "loss": 0.3207,
+      "step": 27899
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.260930096372075e-08,
+      "loss": 0.2909,
+      "step": 27900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.248415038422217e-08,
+      "loss": 0.2968,
+      "step": 27901
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.23591246204791e-08,
+      "loss": 0.3357,
+      "step": 27902
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.22342236740614e-08,
+      "loss": 0.3181,
+      "step": 27903
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.210944754653891e-08,
+      "loss": 0.2712,
+      "step": 27904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.198479623947817e-08,
+      "loss": 0.3258,
+      "step": 27905
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.18602697544457e-08,
+      "loss": 0.3267,
+      "step": 27906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.173586809300358e-08,
+      "loss": 0.312,
+      "step": 27907
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.161159125671612e-08,
+      "loss": 0.2778,
+      "step": 27908
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.148743924714317e-08,
+      "loss": 0.2792,
+      "step": 27909
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.136341206584572e-08,
+      "loss": 0.3115,
+      "step": 27910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.123950971438031e-08,
+      "loss": 0.2985,
+      "step": 27911
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.111573219430234e-08,
+      "loss": 0.3078,
+      "step": 27912
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.099207950716724e-08,
+      "loss": 0.3084,
+      "step": 27913
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.086855165452932e-08,
+      "loss": 0.3233,
+      "step": 27914
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.074514863793846e-08,
+      "loss": 0.3026,
+      "step": 27915
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.062187045894563e-08,
+      "loss": 0.3103,
+      "step": 27916
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.04987171190996e-08,
+      "loss": 0.3092,
+      "step": 27917
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.037568861994469e-08,
+      "loss": 0.3275,
+      "step": 27918
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.025278496302855e-08,
+      "loss": 0.3172,
+      "step": 27919
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.013000614989328e-08,
+      "loss": 0.3206,
+      "step": 27920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.00073521820832e-08,
+      "loss": 0.3296,
+      "step": 27921
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.988482306113597e-08,
+      "loss": 0.2888,
+      "step": 27922
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.976241878859368e-08,
+      "loss": 0.3361,
+      "step": 27923
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.964013936599067e-08,
+      "loss": 0.3002,
+      "step": 27924
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.951798479486459e-08,
+      "loss": 0.3087,
+      "step": 27925
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9395955076750886e-08,
+      "loss": 0.2809,
+      "step": 27926
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.927405021317945e-08,
+      "loss": 0.3292,
+      "step": 27927
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.915227020568237e-08,
+      "loss": 0.3014,
+      "step": 27928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9030615055790666e-08,
+      "loss": 0.3108,
+      "step": 27929
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8909084765029764e-08,
+      "loss": 0.3285,
+      "step": 27930
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.878767933492846e-08,
+      "loss": 0.316,
+      "step": 27931
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.866639876700997e-08,
+      "loss": 0.3098,
+      "step": 27932
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8545243062798626e-08,
+      "loss": 0.3247,
+      "step": 27933
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.842421222381545e-08,
+      "loss": 0.3005,
+      "step": 27934
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8303306251581426e-08,
+      "loss": 0.3369,
+      "step": 27935
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.818252514761535e-08,
+      "loss": 0.3333,
+      "step": 27936
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8061868913432686e-08,
+      "loss": 0.319,
+      "step": 27937
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.794133755054998e-08,
+      "loss": 0.3364,
+      "step": 27938
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.782093106048159e-08,
+      "loss": 0.3024,
+      "step": 27939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7700649444737414e-08,
+      "loss": 0.3043,
+      "step": 27940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.758049270483068e-08,
+      "loss": 0.2718,
+      "step": 27941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7460460842270194e-08,
+      "loss": 0.3116,
+      "step": 27942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.734055385856363e-08,
+      "loss": 0.3159,
+      "step": 27943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.722077175521534e-08,
+      "loss": 0.3067,
+      "step": 27944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7101114533731904e-08,
+      "loss": 0.318,
+      "step": 27945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.698158219561434e-08,
+      "loss": 0.3002,
+      "step": 27946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6862174742365886e-08,
+      "loss": 0.3081,
+      "step": 27947
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.674289217548645e-08,
+      "loss": 0.3293,
+      "step": 27948
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.662373449647151e-08,
+      "loss": 0.3221,
+      "step": 27949
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.650470170681876e-08,
+      "loss": 0.3322,
+      "step": 27950
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.638579380802589e-08,
+      "loss": 0.3223,
+      "step": 27951
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6267010801582814e-08,
+      "loss": 0.3442,
+      "step": 27952
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6148352688982774e-08,
+      "loss": 0.3001,
+      "step": 27953
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.602981947171793e-08,
+      "loss": 0.3161,
+      "step": 27954
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5911411151273744e-08,
+      "loss": 0.3143,
+      "step": 27955
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.579312772914014e-08,
+      "loss": 0.2702,
+      "step": 27956
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.56749692068026e-08,
+      "loss": 0.2856,
+      "step": 27957
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.555693558574326e-08,
+      "loss": 0.3024,
+      "step": 27958
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.54390268674454e-08,
+      "loss": 0.3149,
+      "step": 27959
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.532124305339115e-08,
+      "loss": 0.3012,
+      "step": 27960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.520358414505933e-08,
+      "loss": 0.2854,
+      "step": 27961
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.508605014392654e-08,
+      "loss": 0.331,
+      "step": 27962
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.496864105147048e-08,
+      "loss": 0.3181,
+      "step": 27963
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.485135686916443e-08,
+      "loss": 0.3109,
+      "step": 27964
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4734197598482755e-08,
+      "loss": 0.2809,
+      "step": 27965
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.461716324089761e-08,
+      "loss": 0.3162,
+      "step": 27966
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.450025379787671e-08,
+      "loss": 0.3096,
+      "step": 27967
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4383469270889996e-08,
+      "loss": 0.3099,
+      "step": 27968
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4266809661404076e-08,
+      "loss": 0.2887,
+      "step": 27969
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.415027497088443e-08,
+      "loss": 0.316,
+      "step": 27970
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4033865200793236e-08,
+      "loss": 0.2808,
+      "step": 27971
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3917580352593755e-08,
+      "loss": 0.3053,
+      "step": 27972
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3801420427748166e-08,
+      "loss": 0.3128,
+      "step": 27973
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3685385427711954e-08,
+      "loss": 0.3164,
+      "step": 27974
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3569475353945076e-08,
+      "loss": 0.2686,
+      "step": 27975
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.345369020790303e-08,
+      "loss": 0.3457,
+      "step": 27976
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.33380299910391e-08,
+      "loss": 0.3088,
+      "step": 27977
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.322249470480656e-08,
+      "loss": 0.2964,
+      "step": 27978
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.310708435065759e-08,
+      "loss": 0.33,
+      "step": 27979
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.299179893003992e-08,
+      "loss": 0.3216,
+      "step": 27980
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.287663844440238e-08,
+      "loss": 0.3297,
+      "step": 27981
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.276160289519161e-08,
+      "loss": 0.3008,
+      "step": 27982
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2646692283852e-08,
+      "loss": 0.282,
+      "step": 27983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2531906611826833e-08,
+      "loss": 0.3157,
+      "step": 27984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.24172458805583e-08,
+      "loss": 0.3368,
+      "step": 27985
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.230271009148635e-08,
+      "loss": 0.319,
+      "step": 27986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2188299246048734e-08,
+      "loss": 0.3018,
+      "step": 27987
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.207401334568429e-08,
+      "loss": 0.2908,
+      "step": 27988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.195985239182744e-08,
+      "loss": 0.2862,
+      "step": 27989
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.184581638591146e-08,
+      "loss": 0.3142,
+      "step": 27990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.173190532936856e-08,
+      "loss": 0.3013,
+      "step": 27991
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.161811922363091e-08,
+      "loss": 0.2929,
+      "step": 27992
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.150445807012627e-08,
+      "loss": 0.2873,
+      "step": 27993
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.13909218702835e-08,
+      "loss": 0.3241,
+      "step": 27994
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.12775106255281e-08,
+      "loss": 0.3152,
+      "step": 27995
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.116422433728341e-08,
+      "loss": 0.2749,
+      "step": 27996
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1051063006974934e-08,
+      "loss": 0.3027,
+      "step": 27997
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0938026636020435e-08,
+      "loss": 0.3015,
+      "step": 27998
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0825115225842106e-08,
+      "loss": 0.313,
+      "step": 27999
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.07123287778577e-08,
+      "loss": 0.258,
+      "step": 28000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0599667293483866e-08,
+      "loss": 0.3265,
+      "step": 28001
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.048713077413503e-08,
+      "loss": 0.2989,
+      "step": 28002
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.037471922122561e-08,
+      "loss": 0.3058,
+      "step": 28003
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0262432636166705e-08,
+      "loss": 0.3326,
+      "step": 28004
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.015027102036829e-08,
+      "loss": 0.3371,
+      "step": 28005
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0038234375240356e-08,
+      "loss": 0.3359,
+      "step": 28006
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9926322702189556e-08,
+      "loss": 0.3284,
+      "step": 28007
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9814536002621425e-08,
+      "loss": 0.2904,
+      "step": 28008
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9702874277940406e-08,
+      "loss": 0.2941,
+      "step": 28009
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.959133752954759e-08,
+      "loss": 0.2545,
+      "step": 28010
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9479925758845194e-08,
+      "loss": 0.3063,
+      "step": 28011
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.93686389672321e-08,
+      "loss": 0.3281,
+      "step": 28012
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.925747715610718e-08,
+      "loss": 0.3482,
+      "step": 28013
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.914644032686378e-08,
+      "loss": 0.337,
+      "step": 28014
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.903552848089965e-08,
+      "loss": 0.3023,
+      "step": 28015
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.892474161960703e-08,
+      "loss": 0.3356,
+      "step": 28016
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.881407974437591e-08,
+      "loss": 0.326,
+      "step": 28017
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8703542856597395e-08,
+      "loss": 0.2934,
+      "step": 28018
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8593130957660386e-08,
+      "loss": 0.2992,
+      "step": 28019
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8482844048950426e-08,
+      "loss": 0.3492,
+      "step": 28020
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.83726821318542e-08,
+      "loss": 0.3278,
+      "step": 28021
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8262645207753924e-08,
+      "loss": 0.2808,
+      "step": 28022
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.815273327803183e-08,
+      "loss": 0.3165,
+      "step": 28023
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.804294634406792e-08,
+      "loss": 0.3185,
+      "step": 28024
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.793328440724332e-08,
+      "loss": 0.3107,
+      "step": 28025
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.782374746893359e-08,
+      "loss": 0.304,
+      "step": 28026
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.771433553051541e-08,
+      "loss": 0.2949,
+      "step": 28027
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.760504859336212e-08,
+      "loss": 0.3046,
+      "step": 28028
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.749588665884708e-08,
+      "loss": 0.309,
+      "step": 28029
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.738684972834029e-08,
+      "loss": 0.3317,
+      "step": 28030
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.727793780321288e-08,
+      "loss": 0.2861,
+      "step": 28031
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.716915088483265e-08,
+      "loss": 0.2785,
+      "step": 28032
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.706048897456406e-08,
+      "loss": 0.3077,
+      "step": 28033
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.695195207377379e-08,
+      "loss": 0.3193,
+      "step": 28034
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.684354018382409e-08,
+      "loss": 0.2719,
+      "step": 28035
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.67352533060772e-08,
+      "loss": 0.3045,
+      "step": 28036
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.662709144189315e-08,
+      "loss": 0.2938,
+      "step": 28037
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.651905459262973e-08,
+      "loss": 0.3178,
+      "step": 28038
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6411142759644753e-08,
+      "loss": 0.3088,
+      "step": 28039
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.63033559442938e-08,
+      "loss": 0.2966,
+      "step": 28040
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6195694147929123e-08,
+      "loss": 0.269,
+      "step": 28041
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.60881573719052e-08,
+      "loss": 0.2967,
+      "step": 28042
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.598074561757093e-08,
+      "loss": 0.3024,
+      "step": 28043
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.587345888627526e-08,
+      "loss": 0.2862,
+      "step": 28044
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5766297179367094e-08,
+      "loss": 0.3052,
+      "step": 28045
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5659260498192025e-08,
+      "loss": 0.3078,
+      "step": 28046
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5552348844093434e-08,
+      "loss": 0.2863,
+      "step": 28047
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5445562218415785e-08,
+      "loss": 0.3058,
+      "step": 28048
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5338900622498014e-08,
+      "loss": 0.289,
+      "step": 28049
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.523236405768239e-08,
+      "loss": 0.2954,
+      "step": 28050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.51259525253056e-08,
+      "loss": 0.3602,
+      "step": 28051
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5019666026704376e-08,
+      "loss": 0.2977,
+      "step": 28052
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4913504563214304e-08,
+      "loss": 0.3134,
+      "step": 28053
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4807468136166544e-08,
+      "loss": 0.2845,
+      "step": 28054
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.470155674689558e-08,
+      "loss": 0.3225,
+      "step": 28055
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.459577039673035e-08,
+      "loss": 0.3462,
+      "step": 28056
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.44901090869998e-08,
+      "loss": 0.3284,
+      "step": 28057
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.438457281903063e-08,
+      "loss": 0.317,
+      "step": 28058
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.427916159414958e-08,
+      "loss": 0.3171,
+      "step": 28059
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.417387541368001e-08,
+      "loss": 0.3071,
+      "step": 28060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.40687142789431e-08,
+      "loss": 0.3298,
+      "step": 28061
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3963678191262236e-08,
+      "loss": 0.3119,
+      "step": 28062
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.385876715195414e-08,
+      "loss": 0.3245,
+      "step": 28063
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3753981162337755e-08,
+      "loss": 0.3382,
+      "step": 28064
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.36493202237287e-08,
+      "loss": 0.2851,
+      "step": 28065
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.354478433744258e-08,
+      "loss": 0.3052,
+      "step": 28066
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3440373504790576e-08,
+      "loss": 0.3032,
+      "step": 28067
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3336087727086084e-08,
+      "loss": 0.2872,
+      "step": 28068
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3231927005636944e-08,
+      "loss": 0.3145,
+      "step": 28069
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.31278913417521e-08,
+      "loss": 0.2737,
+      "step": 28070
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3023980736740525e-08,
+      "loss": 0.314,
+      "step": 28071
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.292019519190338e-08,
+      "loss": 0.3174,
+      "step": 28072
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.281653470854741e-08,
+      "loss": 0.3198,
+      "step": 28073
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.271299928797379e-08,
+      "loss": 0.3428,
+      "step": 28074
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.260958893148148e-08,
+      "loss": 0.3215,
+      "step": 28075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2506303640370564e-08,
+      "loss": 0.2808,
+      "step": 28076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.240314341593887e-08,
+      "loss": 0.2836,
+      "step": 28077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.230010825948094e-08,
+      "loss": 0.2732,
+      "step": 28078
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.219719817229129e-08,
+      "loss": 0.3303,
+      "step": 28079
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2094413155662205e-08,
+      "loss": 0.3241,
+      "step": 28080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1991753210886e-08,
+      "loss": 0.3276,
+      "step": 28081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.188921833925053e-08,
+      "loss": 0.3133,
+      "step": 28082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.178680854204475e-08,
+      "loss": 0.2735,
+      "step": 28083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.168452382055321e-08,
+      "loss": 0.281,
+      "step": 28084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.158236417606265e-08,
+      "loss": 0.2963,
+      "step": 28085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.148032960985537e-08,
+      "loss": 0.3206,
+      "step": 28086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.137842012321258e-08,
+      "loss": 0.3201,
+      "step": 28087
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.127663571741436e-08,
+      "loss": 0.3099,
+      "step": 28088
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.117497639373969e-08,
+      "loss": 0.3145,
+      "step": 28089
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.107344215346421e-08,
+      "loss": 0.3028,
+      "step": 28090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0972032997864675e-08,
+      "loss": 0.31,
+      "step": 28091
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.087074892821452e-08,
+      "loss": 0.2939,
+      "step": 28092
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.076958994578495e-08,
+      "loss": 0.2782,
+      "step": 28093
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.066855605184827e-08,
+      "loss": 0.3304,
+      "step": 28094
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.056764724767126e-08,
+      "loss": 0.302,
+      "step": 28095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0466863534522895e-08,
+      "loss": 0.2951,
+      "step": 28096
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.036620491366772e-08,
+      "loss": 0.2816,
+      "step": 28097
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.026567138637139e-08,
+      "loss": 0.3148,
+      "step": 28098
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0165262953896225e-08,
+      "loss": 0.2974,
+      "step": 28099
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.006497961750344e-08,
+      "loss": 0.2794,
+      "step": 28100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9964821378450926e-08,
+      "loss": 0.293,
+      "step": 28101
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9864788237999885e-08,
+      "loss": 0.2959,
+      "step": 28102
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.976488019740377e-08,
+      "loss": 0.2733,
+      "step": 28103
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.966509725791823e-08,
+      "loss": 0.2862,
+      "step": 28104
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9565439420797824e-08,
+      "loss": 0.3099,
+      "step": 28105
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.946590668729378e-08,
+      "loss": 0.2963,
+      "step": 28106
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9366499058653975e-08,
+      "loss": 0.314,
+      "step": 28107
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.926721653613075e-08,
+      "loss": 0.2969,
+      "step": 28108
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.916805912096755e-08,
+      "loss": 0.3033,
+      "step": 28109
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.906902681441227e-08,
+      "loss": 0.3196,
+      "step": 28110
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.897011961770836e-08,
+      "loss": 0.2979,
+      "step": 28111
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887133753209704e-08,
+      "loss": 0.2845,
+      "step": 28112
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8772680558819554e-08,
+      "loss": 0.2683,
+      "step": 28113
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.86741486991149e-08,
+      "loss": 0.2846,
+      "step": 28114
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.857574195422098e-08,
+      "loss": 0.3158,
+      "step": 28115
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8477460325373475e-08,
+      "loss": 0.3058,
+      "step": 28116
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8379303813806945e-08,
+      "loss": 0.3099,
+      "step": 28117
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.828127242075485e-08,
+      "loss": 0.3346,
+      "step": 28118
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.818336614744844e-08,
+      "loss": 0.3174,
+      "step": 28119
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.808558499511561e-08,
+      "loss": 0.3093,
+      "step": 28120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7987928964986487e-08,
+      "loss": 0.2841,
+      "step": 28121
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.789039805828565e-08,
+      "loss": 0.3027,
+      "step": 28122
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7792992276241006e-08,
+      "loss": 0.3332,
+      "step": 28123
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.76957116200738e-08,
+      "loss": 0.3411,
+      "step": 28124
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.759855609100527e-08,
+      "loss": 0.2987,
+      "step": 28125
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7501525690258886e-08,
+      "loss": 0.3304,
+      "step": 28126
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7404620419050355e-08,
+      "loss": 0.2795,
+      "step": 28127
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7307840278597574e-08,
+      "loss": 0.3194,
+      "step": 28128
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7211185270116244e-08,
+      "loss": 0.3047,
+      "step": 28129
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.711465539481984e-08,
+      "loss": 0.3082,
+      "step": 28130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.701825065392184e-08,
+      "loss": 0.2873,
+      "step": 28131
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.692197104863349e-08,
+      "loss": 0.2905,
+      "step": 28132
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.682581658016271e-08,
+      "loss": 0.2968,
+      "step": 28133
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.672978724971632e-08,
+      "loss": 0.29,
+      "step": 28134
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.663388305850335e-08,
+      "loss": 0.2879,
+      "step": 28135
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.653810400772617e-08,
+      "loss": 0.3199,
+      "step": 28136
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6442450098589375e-08,
+      "loss": 0.3352,
+      "step": 28137
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6346921332292004e-08,
+      "loss": 0.2932,
+      "step": 28138
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.625151771003643e-08,
+      "loss": 0.2539,
+      "step": 28139
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6156239233019475e-08,
+      "loss": 0.2721,
+      "step": 28140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.606108590243795e-08,
+      "loss": 0.2776,
+      "step": 28141
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.596605771948758e-08,
+      "loss": 0.3085,
+      "step": 28142
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5871154685360735e-08,
+      "loss": 0.3025,
+      "step": 28143
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.57763768012509e-08,
+      "loss": 0.2913,
+      "step": 28144
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5681724068348246e-08,
+      "loss": 0.3164,
+      "step": 28145
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.55871964878407e-08,
+      "loss": 0.3026,
+      "step": 28146
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5492794060916213e-08,
+      "loss": 0.2953,
+      "step": 28147
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.53985167887605e-08,
+      "loss": 0.3274,
+      "step": 28148
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5304364672557044e-08,
+      "loss": 0.2886,
+      "step": 28149
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.521033771348936e-08,
+      "loss": 0.2722,
+      "step": 28150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5116435912737615e-08,
+      "loss": 0.3116,
+      "step": 28151
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.502265927148085e-08,
+      "loss": 0.3006,
+      "step": 28152
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.492900779089814e-08,
+      "loss": 0.3033,
+      "step": 28153
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.48354814721652e-08,
+      "loss": 0.2836,
+      "step": 28154
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.474208031645776e-08,
+      "loss": 0.2821,
+      "step": 28155
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4648804324946e-08,
+      "loss": 0.2767,
+      "step": 28156
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.455565349880452e-08,
+      "loss": 0.3106,
+      "step": 28157
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.44626278392024e-08,
+      "loss": 0.3171,
+      "step": 28158
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4369727347307594e-08,
+      "loss": 0.2805,
+      "step": 28159
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.427695202428694e-08,
+      "loss": 0.3263,
+      "step": 28160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.418430187130617e-08,
+      "loss": 0.3144,
+      "step": 28161
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.40917768895277e-08,
+      "loss": 0.3051,
+      "step": 28162
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3999377080115026e-08,
+      "loss": 0.3031,
+      "step": 28163
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.390710244422945e-08,
+      "loss": 0.3202,
+      "step": 28164
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.381495298302784e-08,
+      "loss": 0.2926,
+      "step": 28165
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.372292869766924e-08,
+      "loss": 0.3392,
+      "step": 28166
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.363102958930831e-08,
+      "loss": 0.3466,
+      "step": 28167
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3539255659100767e-08,
+      "loss": 0.2934,
+      "step": 28168
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3447606908196815e-08,
+      "loss": 0.3229,
+      "step": 28169
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3356083337749975e-08,
+      "loss": 0.2852,
+      "step": 28170
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.326468494890933e-08,
+      "loss": 0.3382,
+      "step": 28171
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.317341174282174e-08,
+      "loss": 0.3211,
+      "step": 28172
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.308226372063406e-08,
+      "loss": 0.3146,
+      "step": 28173
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2991240883492036e-08,
+      "loss": 0.3024,
+      "step": 28174
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.29003432325381e-08,
+      "loss": 0.2759,
+      "step": 28175
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2809570768913556e-08,
+      "loss": 0.2769,
+      "step": 28176
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2718923493759714e-08,
+      "loss": 0.3276,
+      "step": 28177
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.262840140821455e-08,
+      "loss": 0.2924,
+      "step": 28178
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2538004513413825e-08,
+      "loss": 0.3159,
+      "step": 28179
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.24477328104944e-08,
+      "loss": 0.3002,
+      "step": 28180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.235758630058983e-08,
+      "loss": 0.3257,
+      "step": 28181
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2267564984832524e-08,
+      "loss": 0.312,
+      "step": 28182
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.21776688643527e-08,
+      "loss": 0.2782,
+      "step": 28183
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.208789794027944e-08,
+      "loss": 0.2898,
+      "step": 28184
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.199825221374076e-08,
+      "loss": 0.3029,
+      "step": 28185
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.190873168586239e-08,
+      "loss": 0.3166,
+      "step": 28186
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1819336357767907e-08,
+      "loss": 0.2812,
+      "step": 28187
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1730066230581944e-08,
+      "loss": 0.2972,
+      "step": 28188
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.164092130542362e-08,
+      "loss": 0.2894,
+      "step": 28189
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.155190158341426e-08,
+      "loss": 0.3123,
+      "step": 28190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1463007065670737e-08,
+      "loss": 0.3268,
+      "step": 28191
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1374237753311054e-08,
+      "loss": 0.3028,
+      "step": 28192
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.128559364744876e-08,
+      "loss": 0.2579,
+      "step": 28193
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.119707474919742e-08,
+      "loss": 0.2817,
+      "step": 28194
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1108681059669466e-08,
+      "loss": 0.3087,
+      "step": 28195
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1020412579974015e-08,
+      "loss": 0.3111,
+      "step": 28196
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0932269311221284e-08,
+      "loss": 0.2472,
+      "step": 28197
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.084425125451818e-08,
+      "loss": 0.3306,
+      "step": 28198
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0756358410968246e-08,
+      "loss": 0.3322,
+      "step": 28199
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.066859078167839e-08,
+      "loss": 0.2871,
+      "step": 28200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.058094836774772e-08,
+      "loss": 0.2823,
+      "step": 28201
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.049343117027981e-08,
+      "loss": 0.2931,
+      "step": 28202
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.040603919037155e-08,
+      "loss": 0.3397,
+      "step": 28203
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.031877242912318e-08,
+      "loss": 0.3037,
+      "step": 28204
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.023163088762826e-08,
+      "loss": 0.3007,
+      "step": 28205
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.01446145669837e-08,
+      "loss": 0.3107,
+      "step": 28206
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0057723468279734e-08,
+      "loss": 0.3184,
+      "step": 28207
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9970957592609926e-08,
+      "loss": 0.3241,
+      "step": 28208
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.988431694106342e-08,
+      "loss": 0.3217,
+      "step": 28209
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.979780151472822e-08,
+      "loss": 0.266,
+      "step": 28210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.971141131469124e-08,
+      "loss": 0.3206,
+      "step": 28211
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.962514634203606e-08,
+      "loss": 0.2841,
+      "step": 28212
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9539006597849585e-08,
+      "loss": 0.3176,
+      "step": 28213
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.945299208320984e-08,
+      "loss": 0.2963,
+      "step": 28214
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.93671027991993e-08,
+      "loss": 0.2912,
+      "step": 28215
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.928133874689709e-08,
+      "loss": 0.3291,
+      "step": 28216
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.919569992737792e-08,
+      "loss": 0.2817,
+      "step": 28217
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9110186341720914e-08,
+      "loss": 0.3093,
+      "step": 28218
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9024797990997445e-08,
+      "loss": 0.3101,
+      "step": 28219
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8939534876281094e-08,
+      "loss": 0.288,
+      "step": 28220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.885439699864323e-08,
+      "loss": 0.3297,
+      "step": 28221
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8769384359152995e-08,
+      "loss": 0.2996,
+      "step": 28222
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8684496958876207e-08,
+      "loss": 0.3439,
+      "step": 28223
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8599734798882005e-08,
+      "loss": 0.3122,
+      "step": 28224
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.851509788023288e-08,
+      "loss": 0.2908,
+      "step": 28225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8430586203992416e-08,
+      "loss": 0.2749,
+      "step": 28226
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8346199771221995e-08,
+      "loss": 0.2571,
+      "step": 28227
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8261938582982985e-08,
+      "loss": 0.3052,
+      "step": 28228
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8177802640331208e-08,
+      "loss": 0.2974,
+      "step": 28229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.809379194432582e-08,
+      "loss": 0.3269,
+      "step": 28230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8009906496020423e-08,
+      "loss": 0.2848,
+      "step": 28231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7926146296469726e-08,
+      "loss": 0.2725,
+      "step": 28232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7842511346724e-08,
+      "loss": 0.3347,
+      "step": 28233
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.775900164783463e-08,
+      "loss": 0.2767,
+      "step": 28234
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.767561720085188e-08,
+      "loss": 0.3373,
+      "step": 28235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7592358006821585e-08,
+      "loss": 0.3352,
+      "step": 28236
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7509224066789574e-08,
+      "loss": 0.2687,
+      "step": 28237
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.742621538180057e-08,
+      "loss": 0.3174,
+      "step": 28238
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7343331952895957e-08,
+      "loss": 0.32,
+      "step": 28239
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7260573781118238e-08,
+      "loss": 0.2929,
+      "step": 28240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.717794086750658e-08,
+      "loss": 0.3248,
+      "step": 28241
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7095433213097933e-08,
+      "loss": 0.3458,
+      "step": 28242
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7013050818929242e-08,
+      "loss": 0.2998,
+      "step": 28243
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.693079368603524e-08,
+      "loss": 0.3326,
+      "step": 28244
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.684866181544954e-08,
+      "loss": 0.2738,
+      "step": 28245
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.676665520820243e-08,
+      "loss": 0.3093,
+      "step": 28246
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.668477386532531e-08,
+      "loss": 0.3266,
+      "step": 28247
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.660301778784624e-08,
+      "loss": 0.3272,
+      "step": 28248
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6521386976791076e-08,
+      "loss": 0.2929,
+      "step": 28249
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6439881433185655e-08,
+      "loss": 0.3056,
+      "step": 28250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6358501158054718e-08,
+      "loss": 0.2848,
+      "step": 28251
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6277246152419668e-08,
+      "loss": 0.3449,
+      "step": 28252
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.61961164173008e-08,
+      "loss": 0.3172,
+      "step": 28253
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.611511195371619e-08,
+      "loss": 0.295,
+      "step": 28254
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.603423276268502e-08,
+      "loss": 0.2964,
+      "step": 28255
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5953478845223147e-08,
+      "loss": 0.2777,
+      "step": 28256
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.58728502023442e-08,
+      "loss": 0.2829,
+      "step": 28257
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5792346835059602e-08,
+      "loss": 0.3043,
+      "step": 28258
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5711968744382975e-08,
+      "loss": 0.3114,
+      "step": 28259
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.563171593132241e-08,
+      "loss": 0.3637,
+      "step": 28260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5551588396885983e-08,
+      "loss": 0.311,
+      "step": 28261
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5471586142080673e-08,
+      "loss": 0.3406,
+      "step": 28262
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5391709167910117e-08,
+      "loss": 0.2904,
+      "step": 28263
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5311957475379068e-08,
+      "loss": 0.3295,
+      "step": 28264
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5232331065487836e-08,
+      "loss": 0.3221,
+      "step": 28265
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5152829939237842e-08,
+      "loss": 0.3159,
+      "step": 28266
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.507345409762607e-08,
+      "loss": 0.2986,
+      "step": 28267
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4994203541650608e-08,
+      "loss": 0.2944,
+      "step": 28268
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4915078272307325e-08,
+      "loss": 0.2923,
+      "step": 28269
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.483607829058876e-08,
+      "loss": 0.307,
+      "step": 28270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4757203597487455e-08,
+      "loss": 0.2848,
+      "step": 28271
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4678454193994838e-08,
+      "loss": 0.2737,
+      "step": 28272
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4599830081099008e-08,
+      "loss": 0.3173,
+      "step": 28273
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.452133125978806e-08,
+      "loss": 0.3071,
+      "step": 28274
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.444295773104899e-08,
+      "loss": 0.2939,
+      "step": 28275
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4364709495864337e-08,
+      "loss": 0.301,
+      "step": 28276
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4286586555216652e-08,
+      "loss": 0.3344,
+      "step": 28277
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4208588910089593e-08,
+      "loss": 0.3318,
+      "step": 28278
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4130716561460155e-08,
+      "loss": 0.3191,
+      "step": 28279
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4052969510308665e-08,
+      "loss": 0.3241,
+      "step": 28280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.39753477576099e-08,
+      "loss": 0.3105,
+      "step": 28281
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3897851304339747e-08,
+      "loss": 0.2502,
+      "step": 28282
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.382048015147187e-08,
+      "loss": 0.2938,
+      "step": 28283
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3743234299976604e-08,
+      "loss": 0.2844,
+      "step": 28284
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.366611375082539e-08,
+      "loss": 0.3245,
+      "step": 28285
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3589118504987462e-08,
+      "loss": 0.3292,
+      "step": 28286
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3512248563427598e-08,
+      "loss": 0.3214,
+      "step": 28287
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.34355039271128e-08,
+      "loss": 0.3145,
+      "step": 28288
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3358884597006748e-08,
+      "loss": 0.2959,
+      "step": 28289
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.328239057407311e-08,
+      "loss": 0.3023,
+      "step": 28290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3206021859270012e-08,
+      "loss": 0.2926,
+      "step": 28291
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3129778453558904e-08,
+      "loss": 0.3025,
+      "step": 28292
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.305366035789569e-08,
+      "loss": 0.2986,
+      "step": 28293
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2977667573237384e-08,
+      "loss": 0.3358,
+      "step": 28294
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2901800100539884e-08,
+      "loss": 0.3121,
+      "step": 28295
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2826057940753543e-08,
+      "loss": 0.3229,
+      "step": 28296
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2750441094829822e-08,
+      "loss": 0.3024,
+      "step": 28297
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2674949563720183e-08,
+      "loss": 0.2693,
+      "step": 28298
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2599583348371645e-08,
+      "loss": 0.3173,
+      "step": 28299
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.252434244973123e-08,
+      "loss": 0.2943,
+      "step": 28300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2449226868743736e-08,
+      "loss": 0.3257,
+      "step": 28301
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.237423660635396e-08,
+      "loss": 0.3209,
+      "step": 28302
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2299371663501156e-08,
+      "loss": 0.3225,
+      "step": 28303
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2224632041127904e-08,
+      "loss": 0.3246,
+      "step": 28304
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2150017740171226e-08,
+      "loss": 0.2973,
+      "step": 28305
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2075528761570374e-08,
+      "loss": 0.286,
+      "step": 28306
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.200116510625905e-08,
+      "loss": 0.3437,
+      "step": 28307
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.192692677517094e-08,
+      "loss": 0.2697,
+      "step": 28308
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1852813769240866e-08,
+      "loss": 0.2949,
+      "step": 28309
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1778826089398074e-08,
+      "loss": 0.2979,
+      "step": 28310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1704963736571827e-08,
+      "loss": 0.3137,
+      "step": 28311
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1631226711690267e-08,
+      "loss": 0.3249,
+      "step": 28312
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1557615015678212e-08,
+      "loss": 0.281,
+      "step": 28313
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1484128649462698e-08,
+      "loss": 0.3122,
+      "step": 28314
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1410767613965212e-08,
+      "loss": 0.2791,
+      "step": 28315
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1337531910107234e-08,
+      "loss": 0.3134,
+      "step": 28316
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1264421538808033e-08,
+      "loss": 0.3097,
+      "step": 28317
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1191436500987984e-08,
+      "loss": 0.2933,
+      "step": 28318
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1118576797560796e-08,
+      "loss": 0.3074,
+      "step": 28319
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.104584242944352e-08,
+      "loss": 0.283,
+      "step": 28320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0973233397549865e-08,
+      "loss": 0.2947,
+      "step": 28321
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0900749702791322e-08,
+      "loss": 0.2864,
+      "step": 28322
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.082839134607828e-08,
+      "loss": 0.2908,
+      "step": 28323
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.07561583283189e-08,
+      "loss": 0.3424,
+      "step": 28324
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.068405065042245e-08,
+      "loss": 0.3153,
+      "step": 28325
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0612068313291545e-08,
+      "loss": 0.3016,
+      "step": 28326
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0540211317833237e-08,
+      "loss": 0.309,
+      "step": 28327
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0468479664947915e-08,
+      "loss": 0.2874,
+      "step": 28328
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0396873355538195e-08,
+      "loss": 0.3098,
+      "step": 28329
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.032539239050224e-08,
+      "loss": 0.3252,
+      "step": 28330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0254036770738227e-08,
+      "loss": 0.3431,
+      "step": 28331
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0182806497142103e-08,
+      "loss": 0.3092,
+      "step": 28332
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0111701570609822e-08,
+      "loss": 0.3197,
+      "step": 28333
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0040721992032886e-08,
+      "loss": 0.3184,
+      "step": 28334
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.996986776230281e-08,
+      "loss": 0.3116,
+      "step": 28335
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9899138882311098e-08,
+      "loss": 0.3524,
+      "step": 28336
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.982853535294482e-08,
+      "loss": 0.3133,
+      "step": 28337
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.975805717509105e-08,
+      "loss": 0.3148,
+      "step": 28338
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.968770434963463e-08,
+      "loss": 0.2995,
+      "step": 28339
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9617476877460405e-08,
+      "loss": 0.3263,
+      "step": 28340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.95473747594499e-08,
+      "loss": 0.2903,
+      "step": 28341
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9477397996482405e-08,
+      "loss": 0.3317,
+      "step": 28342
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.940754658943833e-08,
+      "loss": 0.3273,
+      "step": 28343
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9337820539193642e-08,
+      "loss": 0.3258,
+      "step": 28344
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9268219846625413e-08,
+      "loss": 0.3283,
+      "step": 28345
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9198744512607392e-08,
+      "loss": 0.2873,
+      "step": 28346
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9129394538012215e-08,
+      "loss": 0.2802,
+      "step": 28347
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.906016992371029e-08,
+      "loss": 0.3063,
+      "step": 28348
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8991070670573154e-08,
+      "loss": 0.2988,
+      "step": 28349
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8922096779465658e-08,
+      "loss": 0.2494,
+      "step": 28350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8853248251256008e-08,
+      "loss": 0.2874,
+      "step": 28351
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8784525086809058e-08,
+      "loss": 0.2919,
+      "step": 28352
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8715927286986347e-08,
+      "loss": 0.2986,
+      "step": 28353
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8647454852651625e-08,
+      "loss": 0.2557,
+      "step": 28354
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8579107784663097e-08,
+      "loss": 0.3342,
+      "step": 28355
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8510886083881186e-08,
+      "loss": 0.3311,
+      "step": 28356
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8442789751160763e-08,
+      "loss": 0.2921,
+      "step": 28357
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8374818787357808e-08,
+      "loss": 0.3017,
+      "step": 28358
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8306973193326083e-08,
+      "loss": 0.3095,
+      "step": 28359
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.823925296991824e-08,
+      "loss": 0.3092,
+      "step": 28360
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8171658117984714e-08,
+      "loss": 0.3173,
+      "step": 28361
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.810418863837371e-08,
+      "loss": 0.3135,
+      "step": 28362
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8036844531932328e-08,
+      "loss": 0.3314,
+      "step": 28363
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7969625799508782e-08,
+      "loss": 0.3007,
+      "step": 28364
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.790253244194462e-08,
+      "loss": 0.2886,
+      "step": 28365
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.783556446008472e-08,
+      "loss": 0.3031,
+      "step": 28366
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7768721854768412e-08,
+      "loss": 0.3013,
+      "step": 28367
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7702004626836135e-08,
+      "loss": 0.3051,
+      "step": 28368
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7635412777126105e-08,
+      "loss": 0.3348,
+      "step": 28369
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.756894630647432e-08,
+      "loss": 0.2993,
+      "step": 28370
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7502605215715672e-08,
+      "loss": 0.2989,
+      "step": 28371
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7436389505682828e-08,
+      "loss": 0.303,
+      "step": 28372
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.737029917720845e-08,
+      "loss": 0.3192,
+      "step": 28373
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7304334231121878e-08,
+      "loss": 0.2918,
+      "step": 28374
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7238494668252447e-08,
+      "loss": 0.298,
+      "step": 28375
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7172780489426166e-08,
+      "loss": 0.3615,
+      "step": 28376
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.710719169547015e-08,
+      "loss": 0.3005,
+      "step": 28377
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.704172828720596e-08,
+      "loss": 0.3103,
+      "step": 28378
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.697639026545628e-08,
+      "loss": 0.2846,
+      "step": 28379
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6911177631042664e-08,
+      "loss": 0.3235,
+      "step": 28380
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6846090384783354e-08,
+      "loss": 0.3105,
+      "step": 28381
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.678112852749658e-08,
+      "loss": 0.3042,
+      "step": 28382
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6716292059998364e-08,
+      "loss": 0.2876,
+      "step": 28383
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6651580983102488e-08,
+      "loss": 0.2793,
+      "step": 28384
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6586995297621646e-08,
+      "loss": 0.287,
+      "step": 28385
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.65225350043674e-08,
+      "loss": 0.3356,
+      "step": 28386
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6458200104149115e-08,
+      "loss": 0.3021,
+      "step": 28387
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.639399059777502e-08,
+      "loss": 0.2954,
+      "step": 28388
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.632990648605115e-08,
+      "loss": 0.3019,
+      "step": 28389
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.626594776978352e-08,
+      "loss": 0.3095,
+      "step": 28390
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.620211444977482e-08,
+      "loss": 0.3019,
+      "step": 28391
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6138406526826633e-08,
+      "loss": 0.3,
+      "step": 28392
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.607482400173832e-08,
+      "loss": 0.3094,
+      "step": 28393
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6011366875310353e-08,
+      "loss": 0.3005,
+      "step": 28394
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.594803514833876e-08,
+      "loss": 0.3109,
+      "step": 28395
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.588482882161846e-08,
+      "loss": 0.2916,
+      "step": 28396
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.582174789594437e-08,
+      "loss": 0.298,
+      "step": 28397
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.575879237210809e-08,
+      "loss": 0.3302,
+      "step": 28398
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5695962250900087e-08,
+      "loss": 0.2846,
+      "step": 28399
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5633257533110845e-08,
+      "loss": 0.2916,
+      "step": 28400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5570678219526403e-08,
+      "loss": 0.3055,
+      "step": 28401
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5508224310933904e-08,
+      "loss": 0.3406,
+      "step": 28402
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5445895808117174e-08,
+      "loss": 0.2921,
+      "step": 28403
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.538369271185891e-08,
+      "loss": 0.2909,
+      "step": 28404
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5321615022940716e-08,
+      "loss": 0.3114,
+      "step": 28405
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5259662742141966e-08,
+      "loss": 0.3187,
+      "step": 28406
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.519783587024204e-08,
+      "loss": 0.2816,
+      "step": 28407
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5136134408015867e-08,
+      "loss": 0.2795,
+      "step": 28408
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.507455835623839e-08,
+      "loss": 0.299,
+      "step": 28409
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.501310771568454e-08,
+      "loss": 0.2985,
+      "step": 28410
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4951782487125922e-08,
+      "loss": 0.317,
+      "step": 28411
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4890582671330812e-08,
+      "loss": 0.273,
+      "step": 28412
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4829508269069703e-08,
+      "loss": 0.2935,
+      "step": 28413
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4768559281109762e-08,
+      "loss": 0.3264,
+      "step": 28414
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.470773570821482e-08,
+      "loss": 0.3297,
+      "step": 28415
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4647037551150934e-08,
+      "loss": 0.2844,
+      "step": 28416
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4586464810679712e-08,
+      "loss": 0.3245,
+      "step": 28417
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4526017487560551e-08,
+      "loss": 0.2877,
+      "step": 28418
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4465695582553951e-08,
+      "loss": 0.324,
+      "step": 28419
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4405499096418196e-08,
+      "loss": 0.2701,
+      "step": 28420
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4345428029909347e-08,
+      "loss": 0.3246,
+      "step": 28421
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4285482383780136e-08,
+      "loss": 0.3008,
+      "step": 28422
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4225662158784403e-08,
+      "loss": 0.3073,
+      "step": 28423
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.416596735567377e-08,
+      "loss": 0.3059,
+      "step": 28424
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4106397975197638e-08,
+      "loss": 0.3211,
+      "step": 28425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4046954018105407e-08,
+      "loss": 0.3063,
+      "step": 28426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3987635485142037e-08,
+      "loss": 0.3056,
+      "step": 28427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3928442377052487e-08,
+      "loss": 0.3361,
+      "step": 28428
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3869374694582826e-08,
+      "loss": 0.3149,
+      "step": 28429
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3810432438471354e-08,
+      "loss": 0.284,
+      "step": 28430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3751615609460812e-08,
+      "loss": 0.296,
+      "step": 28431
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3692924208289492e-08,
+      "loss": 0.3392,
+      "step": 28432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3634358235694589e-08,
+      "loss": 0.297,
+      "step": 28433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3575917692411067e-08,
+      "loss": 0.3091,
+      "step": 28434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3517602579172783e-08,
+      "loss": 0.2969,
+      "step": 28435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3459412896713597e-08,
+      "loss": 0.2989,
+      "step": 28436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3401348645762924e-08,
+      "loss": 0.2457,
+      "step": 28437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.334340982705018e-08,
+      "loss": 0.2935,
+      "step": 28438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3285596441304782e-08,
+      "loss": 0.3325,
+      "step": 28439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3227908489250596e-08,
+      "loss": 0.3202,
+      "step": 28440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3170345971613707e-08,
+      "loss": 0.2801,
+      "step": 28441
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3112908889116871e-08,
+      "loss": 0.3038,
+      "step": 28442
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3055597242480622e-08,
+      "loss": 0.2737,
+      "step": 28443
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2998411032425495e-08,
+      "loss": 0.3195,
+      "step": 28444
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2941350259669804e-08,
+      "loss": 0.288,
+      "step": 28445
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2884414924929645e-08,
+      "loss": 0.3022,
+      "step": 28446
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2827605028921108e-08,
+      "loss": 0.3011,
+      "step": 28447
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.277092057235696e-08,
+      "loss": 0.2828,
+      "step": 28448
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.271436155594885e-08,
+      "loss": 0.3091,
+      "step": 28449
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2657927980408436e-08,
+      "loss": 0.2811,
+      "step": 28450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2601619846444035e-08,
+      "loss": 0.3171,
+      "step": 28451
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2545437154762863e-08,
+      "loss": 0.3105,
+      "step": 28452
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.248937990606991e-08,
+      "loss": 0.2913,
+      "step": 28453
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2433448101069057e-08,
+      "loss": 0.3057,
+      "step": 28454
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2377641740464185e-08,
+      "loss": 0.2696,
+      "step": 28455
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2321960824954737e-08,
+      "loss": 0.3236,
+      "step": 28456
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2266405355242373e-08,
+      "loss": 0.2978,
+      "step": 28457
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.221097533202209e-08,
+      "loss": 0.3133,
+      "step": 28458
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2155670755991112e-08,
+      "loss": 0.3291,
+      "step": 28459
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2100491627845545e-08,
+      "loss": 0.2998,
+      "step": 28460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2045437948275952e-08,
+      "loss": 0.2968,
+      "step": 28461
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1990509717976217e-08,
+      "loss": 0.2608,
+      "step": 28462
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1935706937633574e-08,
+      "loss": 0.3325,
+      "step": 28463
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1881029607938577e-08,
+      "loss": 0.2839,
+      "step": 28464
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1826477729577346e-08,
+      "loss": 0.2875,
+      "step": 28465
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1772051303235998e-08,
+      "loss": 0.2536,
+      "step": 28466
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.17177503295951e-08,
+      "loss": 0.3438,
+      "step": 28467
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1663574809339661e-08,
+      "loss": 0.2565,
+      "step": 28468
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1609524743149137e-08,
+      "loss": 0.307,
+      "step": 28469
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1555600131702982e-08,
+      "loss": 0.2836,
+      "step": 28470
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1501800975677324e-08,
+      "loss": 0.309,
+      "step": 28471
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1448127275748289e-08,
+      "loss": 0.3416,
+      "step": 28472
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1394579032589782e-08,
+      "loss": 0.3002,
+      "step": 28473
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1341156246875707e-08,
+      "loss": 0.3073,
+      "step": 28474
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.128785891927553e-08,
+      "loss": 0.2933,
+      "step": 28475
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1234687050459824e-08,
+      "loss": 0.3168,
+      "step": 28476
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1181640641094727e-08,
+      "loss": 0.3224,
+      "step": 28477
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1128719691848588e-08,
+      "loss": 0.3083,
+      "step": 28478
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1075924203385324e-08,
+      "loss": 0.2876,
+      "step": 28479
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1023254176367737e-08,
+      "loss": 0.3217,
+      "step": 28480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.097070961145752e-08,
+      "loss": 0.346,
+      "step": 28481
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0918290509314145e-08,
+      "loss": 0.3059,
+      "step": 28482
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0865996870597084e-08,
+      "loss": 0.3266,
+      "step": 28483
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.081382869596248e-08,
+      "loss": 0.2809,
+      "step": 28484
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0761785986065365e-08,
+      "loss": 0.2859,
+      "step": 28485
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0709868741559659e-08,
+      "loss": 0.2917,
+      "step": 28486
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0658076963098174e-08,
+      "loss": 0.3015,
+      "step": 28487
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.060641065133039e-08,
+      "loss": 0.2734,
+      "step": 28488
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0554869806904677e-08,
+      "loss": 0.303,
+      "step": 28489
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0503454430470517e-08,
+      "loss": 0.3544,
+      "step": 28490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0452164522671837e-08,
+      "loss": 0.2942,
+      "step": 28491
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0401000084152568e-08,
+      "loss": 0.3087,
+      "step": 28492
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.034996111555775e-08,
+      "loss": 0.3009,
+      "step": 28493
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0299047617525759e-08,
+      "loss": 0.3174,
+      "step": 28494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0248259590698306e-08,
+      "loss": 0.3044,
+      "step": 28495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0197597035711548e-08,
+      "loss": 0.307,
+      "step": 28496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0147059953202754e-08,
+      "loss": 0.3204,
+      "step": 28497
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0096648343805859e-08,
+      "loss": 0.3232,
+      "step": 28498
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0046362208154802e-08,
+      "loss": 0.3024,
+      "step": 28499
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.996201546881302e-09,
+      "loss": 0.3307,
+      "step": 28500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.946166360614851e-09,
+      "loss": 0.2981,
+      "step": 28501
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.896256649983838e-09,
+      "loss": 0.297,
+      "step": 28502
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.846472415615538e-09,
+      "loss": 0.2968,
+      "step": 28503
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.796813658135006e-09,
+      "loss": 0.2875,
+      "step": 28504
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.74728037816508e-09,
+      "loss": 0.3292,
+      "step": 28505
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.697872576330813e-09,
+      "loss": 0.3001,
+      "step": 28506
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.64859025324949e-09,
+      "loss": 0.3278,
+      "step": 28507
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.599433409540615e-09,
+      "loss": 0.2742,
+      "step": 28508
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.550402045823692e-09,
+      "loss": 0.3154,
+      "step": 28509
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.501496162712676e-09,
+      "loss": 0.2814,
+      "step": 28510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.452715760822628e-09,
+      "loss": 0.2974,
+      "step": 28511
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.404060840766394e-09,
+      "loss": 0.3207,
+      "step": 28512
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.355531403154595e-09,
+      "loss": 0.2904,
+      "step": 28513
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.307127448595631e-09,
+      "loss": 0.3141,
+      "step": 28514
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.25884897770013e-09,
+      "loss": 0.3266,
+      "step": 28515
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.210695991073159e-09,
+      "loss": 0.3214,
+      "step": 28516
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.162668489318682e-09,
+      "loss": 0.3334,
+      "step": 28517
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.11476647304066e-09,
+      "loss": 0.2856,
+      "step": 28518
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.066989942840832e-09,
+      "loss": 0.3025,
+      "step": 28519
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.01933889931872e-09,
+      "loss": 0.2974,
+      "step": 28520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.971813343073843e-09,
+      "loss": 0.2913,
+      "step": 28521
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.924413274702392e-09,
+      "loss": 0.3157,
+      "step": 28522
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.877138694799448e-09,
+      "loss": 0.2742,
+      "step": 28523
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.829989603958978e-09,
+      "loss": 0.2893,
+      "step": 28524
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.78296600277495e-09,
+      "loss": 0.287,
+      "step": 28525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.736067891835787e-09,
+      "loss": 0.3067,
+      "step": 28526
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.689295271729902e-09,
+      "loss": 0.3176,
+      "step": 28527
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.642648143047939e-09,
+      "loss": 0.3179,
+      "step": 28528
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.59612650637276e-09,
+      "loss": 0.2753,
+      "step": 28529
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.549730362289455e-09,
+      "loss": 0.3276,
+      "step": 28530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.503459711382001e-09,
+      "loss": 0.2894,
+      "step": 28531
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.457314554231045e-09,
+      "loss": 0.3259,
+      "step": 28532
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.411294891415011e-09,
+      "loss": 0.2838,
+      "step": 28533
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.365400723512329e-09,
+      "loss": 0.3087,
+      "step": 28534
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.319632051100313e-09,
+      "loss": 0.2649,
+      "step": 28535
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.273988874754057e-09,
+      "loss": 0.3166,
+      "step": 28536
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.22847119504533e-09,
+      "loss": 0.3161,
+      "step": 28537
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.183079012545891e-09,
+      "loss": 0.3266,
+      "step": 28538
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.13781232782751e-09,
+      "loss": 0.3174,
+      "step": 28539
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.09267114145751e-09,
+      "loss": 0.3049,
+      "step": 28540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.047655454003212e-09,
+      "loss": 0.2698,
+      "step": 28541
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.002765266030832e-09,
+      "loss": 0.2708,
+      "step": 28542
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.958000578102143e-09,
+      "loss": 0.295,
+      "step": 28543
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.913361390781139e-09,
+      "loss": 0.3335,
+      "step": 28544
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.86884770462737e-09,
+      "loss": 0.2749,
+      "step": 28545
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.8244595202015e-09,
+      "loss": 0.3221,
+      "step": 28546
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.78019683805975e-09,
+      "loss": 0.3098,
+      "step": 28547
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.736059658757233e-09,
+      "loss": 0.3158,
+      "step": 28548
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.692047982851281e-09,
+      "loss": 0.3024,
+      "step": 28549
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.648161810891453e-09,
+      "loss": 0.2774,
+      "step": 28550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.60440114343064e-09,
+      "loss": 0.3111,
+      "step": 28551
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.560765981018402e-09,
+      "loss": 0.286,
+      "step": 28552
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517256324202083e-09,
+      "loss": 0.3095,
+      "step": 28553
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.473872173529018e-09,
+      "loss": 0.2916,
+      "step": 28554
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.430613529544329e-09,
+      "loss": 0.3075,
+      "step": 28555
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.387480392789803e-09,
+      "loss": 0.3049,
+      "step": 28556
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.3444727638083415e-09,
+      "loss": 0.2987,
+      "step": 28557
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.301590643140621e-09,
+      "loss": 0.2809,
+      "step": 28558
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.258834031323991e-09,
+      "loss": 0.2614,
+      "step": 28559
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.2162029288969096e-09,
+      "loss": 0.2845,
+      "step": 28560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.173697336392282e-09,
+      "loss": 0.3289,
+      "step": 28561
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.131317254347458e-09,
+      "loss": 0.2926,
+      "step": 28562
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.089062683292014e-09,
+      "loss": 0.314,
+      "step": 28563
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.046933623757746e-09,
+      "loss": 0.3173,
+      "step": 28564
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.00493007627423e-09,
+      "loss": 0.3348,
+      "step": 28565
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.963052041367713e-09,
+      "loss": 0.2918,
+      "step": 28566
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.9212995195655494e-09,
+      "loss": 0.2718,
+      "step": 28567
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.879672511390656e-09,
+      "loss": 0.2956,
+      "step": 28568
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.838171017367057e-09,
+      "loss": 0.301,
+      "step": 28569
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.796795038016557e-09,
+      "loss": 0.3085,
+      "step": 28570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.755544573856521e-09,
+      "loss": 0.3185,
+      "step": 28571
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.714419625407642e-09,
+      "loss": 0.3328,
+      "step": 28572
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.673420193185065e-09,
+      "loss": 0.3016,
+      "step": 28573
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.632546277703933e-09,
+      "loss": 0.305,
+      "step": 28574
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.591797879478279e-09,
+      "loss": 0.3126,
+      "step": 28575
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.5511749990188056e-09,
+      "loss": 0.323,
+      "step": 28576
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.510677636836216e-09,
+      "loss": 0.294,
+      "step": 28577
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.470305793438991e-09,
+      "loss": 0.3157,
+      "step": 28578
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.430059469334504e-09,
+      "loss": 0.3172,
+      "step": 28579
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.389938665029016e-09,
+      "loss": 0.2942,
+      "step": 28580
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.3499433810243485e-09,
+      "loss": 0.3082,
+      "step": 28581
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.310073617824541e-09,
+      "loss": 0.2781,
+      "step": 28582
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.2703293759303065e-09,
+      "loss": 0.3413,
+      "step": 28583
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.2307106558401334e-09,
+      "loss": 0.3261,
+      "step": 28584
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.191217458052512e-09,
+      "loss": 0.3251,
+      "step": 28585
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.151849783062602e-09,
+      "loss": 0.3171,
+      "step": 28586
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.1126076313644535e-09,
+      "loss": 0.2906,
+      "step": 28587
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.073491003452114e-09,
+      "loss": 0.2656,
+      "step": 28588
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.034499899817414e-09,
+      "loss": 0.311,
+      "step": 28589
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.995634320947741e-09,
+      "loss": 0.3113,
+      "step": 28590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.956894267333812e-09,
+      "loss": 0.287,
+      "step": 28591
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.918279739459687e-09,
+      "loss": 0.3308,
+      "step": 28592
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.879790737812752e-09,
+      "loss": 0.3002,
+      "step": 28593
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8414272628748434e-09,
+      "loss": 0.3195,
+      "step": 28594
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.80318931512891e-09,
+      "loss": 0.3156,
+      "step": 28595
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.765076895054566e-09,
+      "loss": 0.3133,
+      "step": 28596
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.72709000313032e-09,
+      "loss": 0.3045,
+      "step": 28597
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.689228639833566e-09,
+      "loss": 0.3351,
+      "step": 28598
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.65149280563948e-09,
+      "loss": 0.3128,
+      "step": 28599
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.613882501023238e-09,
+      "loss": 0.3115,
+      "step": 28600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.576397726455574e-09,
+      "loss": 0.3027,
+      "step": 28601
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.539038482408332e-09,
+      "loss": 0.2708,
+      "step": 28602
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.501804769351138e-09,
+      "loss": 0.3328,
+      "step": 28603
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.4646965877502845e-09,
+      "loss": 0.2843,
+      "step": 28604
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.427713938072066e-09,
+      "loss": 0.304,
+      "step": 28605
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.390856820782775e-09,
+      "loss": 0.3137,
+      "step": 28606
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.354125236343155e-09,
+      "loss": 0.2873,
+      "step": 28607
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.317519185216169e-09,
+      "loss": 0.2793,
+      "step": 28608
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.281038667860338e-09,
+      "loss": 0.3004,
+      "step": 28609
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.244683684735297e-09,
+      "loss": 0.2948,
+      "step": 28610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.208454236296234e-09,
+      "loss": 0.2745,
+      "step": 28611
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.172350322998343e-09,
+      "loss": 0.3238,
+      "step": 28612
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.136371945296814e-09,
+      "loss": 0.3391,
+      "step": 28613
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.100519103641288e-09,
+      "loss": 0.3254,
+      "step": 28614
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.064791798482516e-09,
+      "loss": 0.3339,
+      "step": 28615
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.029190030270137e-09,
+      "loss": 0.3056,
+      "step": 28616
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9937137994515715e-09,
+      "loss": 0.3047,
+      "step": 28617
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9583631064709094e-09,
+      "loss": 0.2843,
+      "step": 28618
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.923137951773349e-09,
+      "loss": 0.3314,
+      "step": 28619
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.88803833580076e-09,
+      "loss": 0.3057,
+      "step": 28620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8530642589950104e-09,
+      "loss": 0.3074,
+      "step": 28621
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.818215721793529e-09,
+      "loss": 0.3065,
+      "step": 28622
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.783492724635963e-09,
+      "loss": 0.289,
+      "step": 28623
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.748895267956411e-09,
+      "loss": 0.3319,
+      "step": 28624
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.714423352191188e-09,
+      "loss": 0.3327,
+      "step": 28625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.680076977772175e-09,
+      "loss": 0.288,
+      "step": 28626
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.645856145132355e-09,
+      "loss": 0.2978,
+      "step": 28627
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.611760854699165e-09,
+      "loss": 0.3191,
+      "step": 28628
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.577791106902263e-09,
+      "loss": 0.2941,
+      "step": 28629
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.543946902169083e-09,
+      "loss": 0.2754,
+      "step": 28630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.51022824092262e-09,
+      "loss": 0.3339,
+      "step": 28631
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.476635123588091e-09,
+      "loss": 0.2811,
+      "step": 28632
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.443167550586269e-09,
+      "loss": 0.3534,
+      "step": 28633
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.409825522337929e-09,
+      "loss": 0.3378,
+      "step": 28634
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.376609039262736e-09,
+      "loss": 0.3257,
+      "step": 28635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.343518101777022e-09,
+      "loss": 0.3288,
+      "step": 28636
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3105527102971225e-09,
+      "loss": 0.2996,
+      "step": 28637
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.277712865234929e-09,
+      "loss": 0.3021,
+      "step": 28638
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.244998567005665e-09,
+      "loss": 0.3051,
+      "step": 28639
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.212409816019003e-09,
+      "loss": 0.2944,
+      "step": 28640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.179946612683506e-09,
+      "loss": 0.3168,
+      "step": 28641
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.147608957408844e-09,
+      "loss": 0.3013,
+      "step": 28642
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.11539685059914e-09,
+      "loss": 0.3024,
+      "step": 28643
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.083310292660736e-09,
+      "loss": 0.29,
+      "step": 28644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0513492839944214e-09,
+      "loss": 0.3028,
+      "step": 28645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.019513825004318e-09,
+      "loss": 0.2859,
+      "step": 28646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.987803916087885e-09,
+      "loss": 0.3518,
+      "step": 28647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.956219557645913e-09,
+      "loss": 0.3121,
+      "step": 28648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9247607500725316e-09,
+      "loss": 0.2839,
+      "step": 28649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.89342749376409e-09,
+      "loss": 0.312,
+      "step": 28650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.862219789113608e-09,
+      "loss": 0.32,
+      "step": 28651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8311376365152145e-09,
+      "loss": 0.3,
+      "step": 28652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.800181036356376e-09,
+      "loss": 0.2812,
+      "step": 28653
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.769349989026783e-09,
+      "loss": 0.3322,
+      "step": 28654
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.738644494913901e-09,
+      "loss": 0.2736,
+      "step": 28655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7080645544040894e-09,
+      "loss": 0.3287,
+      "step": 28656
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6776101678814848e-09,
+      "loss": 0.2952,
+      "step": 28657
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6472813357268933e-09,
+      "loss": 0.2919,
+      "step": 28658
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.617078058322232e-09,
+      "loss": 0.3033,
+      "step": 28659
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5870003360483076e-09,
+      "loss": 0.2865,
+      "step": 28660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.557048169280375e-09,
+      "loss": 0.3178,
+      "step": 28661
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5272215583959103e-09,
+      "loss": 0.2988,
+      "step": 28662
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4975205037690587e-09,
+      "loss": 0.3241,
+      "step": 28663
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.467945005773965e-09,
+      "loss": 0.2787,
+      "step": 28664
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.438495064780334e-09,
+      "loss": 0.2985,
+      "step": 28665
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.40917068116009e-09,
+      "loss": 0.3478,
+      "step": 28666
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.379971855280717e-09,
+      "loss": 0.2798,
+      "step": 28667
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3508985875074784e-09,
+      "loss": 0.3111,
+      "step": 28668
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.321950878207858e-09,
+      "loss": 0.329,
+      "step": 28669
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2931287277437885e-09,
+      "loss": 0.296,
+      "step": 28670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2644321364783125e-09,
+      "loss": 0.3365,
+      "step": 28671
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.235861104772253e-09,
+      "loss": 0.3529,
+      "step": 28672
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2074156329819917e-09,
+      "loss": 0.3139,
+      "step": 28673
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.17909572146724e-09,
+      "loss": 0.2976,
+      "step": 28674
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1509013705832703e-09,
+      "loss": 0.3105,
+      "step": 28675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.122832580684243e-09,
+      "loss": 0.3185,
+      "step": 28676
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0948893521209888e-09,
+      "loss": 0.3145,
+      "step": 28677
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0670716852465588e-09,
+      "loss": 0.2894,
+      "step": 28678
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.039379580409563e-09,
+      "loss": 0.2992,
+      "step": 28679
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0118130379575005e-09,
+      "loss": 0.3176,
+      "step": 28680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9843720582378722e-09,
+      "loss": 0.2819,
+      "step": 28681
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9570566415937364e-09,
+      "loss": 0.3127,
+      "step": 28682
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.929866788369262e-09,
+      "loss": 0.3028,
+      "step": 28683
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9028024989052884e-09,
+      "loss": 0.2736,
+      "step": 28684
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.875863773541543e-09,
+      "loss": 0.2883,
+      "step": 28685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8490506126177543e-09,
+      "loss": 0.3221,
+      "step": 28686
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8223630164692097e-09,
+      "loss": 0.3054,
+      "step": 28687
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7958009854311964e-09,
+      "loss": 0.3318,
+      "step": 28688
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7693645198390016e-09,
+      "loss": 0.3271,
+      "step": 28689
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7430536200223624e-09,
+      "loss": 0.3042,
+      "step": 28690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7168682863132345e-09,
+      "loss": 0.3141,
+      "step": 28691
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6908085190402444e-09,
+      "loss": 0.2903,
+      "step": 28692
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6648743185309077e-09,
+      "loss": 0.2791,
+      "step": 28693
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.63906568511052e-09,
+      "loss": 0.3186,
+      "step": 28694
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6133826191032663e-09,
+      "loss": 0.2878,
+      "step": 28695
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5878251208322214e-09,
+      "loss": 0.2776,
+      "step": 28696
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5623931906182396e-09,
+      "loss": 0.2747,
+      "step": 28697
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5370868287799555e-09,
+      "loss": 0.2916,
+      "step": 28698
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5119060356348922e-09,
+      "loss": 0.3185,
+      "step": 28699
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4868508115016844e-09,
+      "loss": 0.2883,
+      "step": 28700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.461921156693414e-09,
+      "loss": 0.2698,
+      "step": 28701
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.437117071522055e-09,
+      "loss": 0.2952,
+      "step": 28702
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4124385563017993e-09,
+      "loss": 0.3095,
+      "step": 28703
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3878856113401792e-09,
+      "loss": 0.3029,
+      "step": 28704
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3634582369480576e-09,
+      "loss": 0.2883,
+      "step": 28705
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3391564334296348e-09,
+      "loss": 0.2794,
+      "step": 28706
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3149802010913323e-09,
+      "loss": 0.3063,
+      "step": 28707
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.290929540237352e-09,
+      "loss": 0.2947,
+      "step": 28708
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2670044511696742e-09,
+      "loss": 0.3073,
+      "step": 28709
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.243204934186949e-09,
+      "loss": 0.2819,
+      "step": 28710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.219530989590046e-09,
+      "loss": 0.3004,
+      "step": 28711
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1959826176765065e-09,
+      "loss": 0.3062,
+      "step": 28712
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1725598187405383e-09,
+      "loss": 0.3046,
+      "step": 28713
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.149262593077461e-09,
+      "loss": 0.321,
+      "step": 28714
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.126090940980374e-09,
+      "loss": 0.3516,
+      "step": 28715
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.103044862737935e-09,
+      "loss": 0.3163,
+      "step": 28716
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0801243586421326e-09,
+      "loss": 0.3135,
+      "step": 28717
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0573294289805145e-09,
+      "loss": 0.2885,
+      "step": 28718
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.034660074037298e-09,
+      "loss": 0.3151,
+      "step": 28719
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0121162941000304e-09,
+      "loss": 0.2617,
+      "step": 28720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.989698089449599e-09,
+      "loss": 0.2751,
+      "step": 28721
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9674054603691095e-09,
+      "loss": 0.3043,
+      "step": 28722
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.945238407138339e-09,
+      "loss": 0.2859,
+      "step": 28723
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9231969300337327e-09,
+      "loss": 0.2859,
+      "step": 28724
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.901281029335067e-09,
+      "loss": 0.316,
+      "step": 28725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.879490705316567e-09,
+      "loss": 0.3387,
+      "step": 28726
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8578259582502367e-09,
+      "loss": 0.2982,
+      "step": 28727
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.836286788410302e-09,
+      "loss": 0.2947,
+      "step": 28728
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8148731960665466e-09,
+      "loss": 0.2813,
+      "step": 28729
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.793585181488755e-09,
+      "loss": 0.3134,
+      "step": 28730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7724227449422703e-09,
+      "loss": 0.3241,
+      "step": 28731
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7513858866946566e-09,
+      "loss": 0.3464,
+      "step": 28732
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.730474607009036e-09,
+      "loss": 0.3193,
+      "step": 28733
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7096889061496424e-09,
+      "loss": 0.2831,
+      "step": 28734
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6890287843762675e-09,
+      "loss": 0.3186,
+      "step": 28735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6684942419475936e-09,
+      "loss": 0.292,
+      "step": 28736
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6480852791234126e-09,
+      "loss": 0.2654,
+      "step": 28737
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6278018961601861e-09,
+      "loss": 0.3329,
+      "step": 28738
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6076440933099347e-09,
+      "loss": 0.3065,
+      "step": 28739
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5876118708291199e-09,
+      "loss": 0.2758,
+      "step": 28740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5677052289664318e-09,
+      "loss": 0.2907,
+      "step": 28741
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5479241679750012e-09,
+      "loss": 0.3273,
+      "step": 28742
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5282686881001873e-09,
+      "loss": 0.2902,
+      "step": 28743
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5087387895906801e-09,
+      "loss": 0.3209,
+      "step": 28744
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.489334472691839e-09,
+      "loss": 0.3178,
+      "step": 28745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4700557376468027e-09,
+      "loss": 0.331,
+      "step": 28746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4509025846976e-09,
+      "loss": 0.3127,
+      "step": 28747
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.431875014085149e-09,
+      "loss": 0.3061,
+      "step": 28748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4129730260481477e-09,
+      "loss": 0.3145,
+      "step": 28749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3941966208252945e-09,
+      "loss": 0.3144,
+      "step": 28750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3755457986497356e-09,
+      "loss": 0.3572,
+      "step": 28751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.357020559757949e-09,
+      "loss": 0.3069,
+      "step": 28752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3386209043819708e-09,
+      "loss": 0.3152,
+      "step": 28753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.320346832752728e-09,
+      "loss": 0.3129,
+      "step": 28754
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3021983451000364e-09,
+      "loss": 0.3194,
+      "step": 28755
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.284175441652602e-09,
+      "loss": 0.2838,
+      "step": 28756
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2662781226346898e-09,
+      "loss": 0.2964,
+      "step": 28757
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2485063882727856e-09,
+      "loss": 0.304,
+      "step": 28758
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2308602387900437e-09,
+      "loss": 0.2809,
+      "step": 28759
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2133396744073989e-09,
+      "loss": 0.3301,
+      "step": 28760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1959446953446752e-09,
+      "loss": 0.3122,
+      "step": 28761
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1786753018216968e-09,
+      "loss": 0.3134,
+      "step": 28762
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.161531494053847e-09,
+      "loss": 0.3196,
+      "step": 28763
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1445132722576192e-09,
+      "loss": 0.3119,
+      "step": 28764
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1276206366461762e-09,
+      "loss": 0.3108,
+      "step": 28765
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1108535874315707e-09,
+      "loss": 0.292,
+      "step": 28766
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.094212124824745e-09,
+      "loss": 0.3026,
+      "step": 28767
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.077696249034421e-09,
+      "loss": 0.3137,
+      "step": 28768
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0613059602682108e-09,
+      "loss": 0.2989,
+      "step": 28769
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0450412587326153e-09,
+      "loss": 0.3167,
+      "step": 28770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0289021446308057e-09,
+      "loss": 0.2671,
+      "step": 28771
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0128886181659525e-09,
+      "loss": 0.3227,
+      "step": 28772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.970006795401166e-10,
+      "loss": 0.3178,
+      "step": 28773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.812383289509175e-10,
+      "loss": 0.2789,
+      "step": 28774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.656015665981956e-10,
+      "loss": 0.3135,
+      "step": 28775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.500903926762395e-10,
+      "loss": 0.305,
+      "step": 28776
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.347048073826692e-10,
+      "loss": 0.2814,
+      "step": 28777
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.19444810908443e-10,
+      "loss": 0.2853,
+      "step": 28778
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.043104034456296e-10,
+      "loss": 0.3258,
+      "step": 28779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.893015851851872e-10,
+      "loss": 0.2963,
+      "step": 28780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.744183563158537e-10,
+      "loss": 0.319,
+      "step": 28781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.596607170230364e-10,
+      "loss": 0.3263,
+      "step": 28782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.450286674921426e-10,
+      "loss": 0.2889,
+      "step": 28783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.305222079085796e-10,
+      "loss": 0.2695,
+      "step": 28784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.161413384544237e-10,
+      "loss": 0.3131,
+      "step": 28785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.01886059308421e-10,
+      "loss": 0.3046,
+      "step": 28786
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.877563706504277e-10,
+      "loss": 0.3187,
+      "step": 28787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.737522726591895e-10,
+      "loss": 0.2722,
+      "step": 28788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.598737655090116e-10,
+      "loss": 0.3444,
+      "step": 28789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.461208493753091e-10,
+      "loss": 0.323,
+      "step": 28790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.324935244301668e-10,
+      "loss": 0.3249,
+      "step": 28791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.18991790845669e-10,
+      "loss": 0.2994,
+      "step": 28792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.056156487894594e-10,
+      "loss": 0.314,
+      "step": 28793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.923650984314024e-10,
+      "loss": 0.3264,
+      "step": 28794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.792401399369209e-10,
+      "loss": 0.2914,
+      "step": 28795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.662407734714383e-10,
+      "loss": 0.2856,
+      "step": 28796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.533669991981572e-10,
+      "loss": 0.3004,
+      "step": 28797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.406188172780603e-10,
+      "loss": 0.3366,
+      "step": 28798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.279962278721296e-10,
+      "loss": 0.3055,
+      "step": 28799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.15499231138017e-10,
+      "loss": 0.3133,
+      "step": 28800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.031278272344843e-10,
+      "loss": 0.2861,
+      "step": 28801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.908820163147422e-10,
+      "loss": 0.273,
+      "step": 28802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.78761798534222e-10,
+      "loss": 0.2967,
+      "step": 28803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.66767174043914e-10,
+      "loss": 0.2708,
+      "step": 28804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.548981429948085e-10,
+      "loss": 0.286,
+      "step": 28805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.431547055367858e-10,
+      "loss": 0.3069,
+      "step": 28806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.315368618163952e-10,
+      "loss": 0.3152,
+      "step": 28807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.200446119801861e-10,
+      "loss": 0.2641,
+      "step": 28808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.086779561724875e-10,
+      "loss": 0.3007,
+      "step": 28809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.97436894535408e-10,
+      "loss": 0.305,
+      "step": 28810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.863214272110561e-10,
+      "loss": 0.3284,
+      "step": 28811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.753315543382098e-10,
+      "loss": 0.2962,
+      "step": 28812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.6446727605564677e-10,
+      "loss": 0.3112,
+      "step": 28813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.537285924988144e-10,
+      "loss": 0.2953,
+      "step": 28814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.431155038031598e-10,
+      "loss": 0.3402,
+      "step": 28815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.326280101019098e-10,
+      "loss": 0.3003,
+      "step": 28816
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.222661115271809e-10,
+      "loss": 0.3122,
+      "step": 28817
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1202980820886917e-10,
+      "loss": 0.3083,
+      "step": 28818
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0191910027576053e-10,
+      "loss": 0.2604,
+      "step": 28819
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9193398785442036e-10,
+      "loss": 0.2785,
+      "step": 28820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8207447107030395e-10,
+      "loss": 0.3189,
+      "step": 28821
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7234055004775613e-10,
+      "loss": 0.2842,
+      "step": 28822
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6273222490779137e-10,
+      "loss": 0.3228,
+      "step": 28823
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.532494957725341e-10,
+      "loss": 0.2825,
+      "step": 28824
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.438923627607782e-10,
+      "loss": 0.3048,
+      "step": 28825
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.346608259890971e-10,
+      "loss": 0.3372,
+      "step": 28826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2555488557406425e-10,
+      "loss": 0.2738,
+      "step": 28827
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1657454163114277e-10,
+      "loss": 0.2631,
+      "step": 28828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.07719794271355e-10,
+      "loss": 0.2903,
+      "step": 28829
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.989906436068335e-10,
+      "loss": 0.2958,
+      "step": 28830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.903870897463801e-10,
+      "loss": 0.3053,
+      "step": 28831
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.819091327999068e-10,
+      "loss": 0.2983,
+      "step": 28832
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7355677287177475e-10,
+      "loss": 0.3207,
+      "step": 28833
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6533001006745497e-10,
+      "loss": 0.2901,
+      "step": 28834
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5722884449130845e-10,
+      "loss": 0.2979,
+      "step": 28835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.492532762443656e-10,
+      "loss": 0.3315,
+      "step": 28836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4140330542765657e-10,
+      "loss": 0.2945,
+      "step": 28837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.336789321377708e-10,
+      "loss": 0.306,
+      "step": 28838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2608015647351822e-10,
+      "loss": 0.2992,
+      "step": 28839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.186069785292677e-10,
+      "loss": 0.3042,
+      "step": 28840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1125939839938824e-10,
+      "loss": 0.2966,
+      "step": 28841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0403741617713857e-10,
+      "loss": 0.3094,
+      "step": 28842
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9694103195133652e-10,
+      "loss": 0.2938,
+      "step": 28843
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8997024581191016e-10,
+      "loss": 0.3136,
+      "step": 28844
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.831250578465671e-10,
+      "loss": 0.3246,
+      "step": 28845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7640546814190473e-10,
+      "loss": 0.3274,
+      "step": 28846
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6981147678118982e-10,
+      "loss": 0.2887,
+      "step": 28847
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6334308384768903e-10,
+      "loss": 0.2904,
+      "step": 28848
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5700028942244872e-10,
+      "loss": 0.2887,
+      "step": 28849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5078309358540489e-10,
+      "loss": 0.3172,
+      "step": 28850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4469149641538338e-10,
+      "loss": 0.3147,
+      "step": 28851
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3872549798676916e-10,
+      "loss": 0.3106,
+      "step": 28852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.328850983761676e-10,
+      "loss": 0.3192,
+      "step": 28853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.271702976568534e-10,
+      "loss": 0.2971,
+      "step": 28854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.215810959009911e-10,
+      "loss": 0.3191,
+      "step": 28855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.161174931774145e-10,
+      "loss": 0.3128,
+      "step": 28856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1077948955606766e-10,
+      "loss": 0.3038,
+      "step": 28857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0556708510245373e-10,
+      "loss": 0.3222,
+      "step": 28858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0048027988318609e-10,
+      "loss": 0.313,
+      "step": 28859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.55190739626577e-11,
+      "loss": 0.2971,
+      "step": 28860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.068346740193079e-11,
+      "loss": 0.3088,
+      "step": 28861
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.597346026206765e-11,
+      "loss": 0.2983,
+      "step": 28862
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.138905260302033e-11,
+      "loss": 0.2958,
+      "step": 28863
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.693024448141017e-11,
+      "loss": 0.3114,
+      "step": 28864
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259703595274836e-11,
+      "loss": 0.3293,
+      "step": 28865
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.838942707365625e-11,
+      "loss": 0.2968,
+      "step": 28866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.430741789520412e-11,
+      "loss": 0.3052,
+      "step": 28867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.035100846957242e-11,
+      "loss": 0.2995,
+      "step": 28868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.652019884450077e-11,
+      "loss": 0.3156,
+      "step": 28869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.2814989071059417e-11,
+      "loss": 0.3117,
+      "step": 28870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.9235379193657287e-11,
+      "loss": 0.3217,
+      "step": 28871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.578136925892374e-11,
+      "loss": 0.2935,
+      "step": 28872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.245295930793702e-11,
+      "loss": 0.3062,
+      "step": 28873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.925014938399585e-11,
+      "loss": 0.2997,
+      "step": 28874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6172939528178467e-11,
+      "loss": 0.3082,
+      "step": 28875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.322132977712222e-11,
+      "loss": 0.3143,
+      "step": 28876
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0395320168574717e-11,
+      "loss": 0.2932,
+      "step": 28877
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7694910739173298e-11,
+      "loss": 0.3212,
+      "step": 28878
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.512010152222466e-11,
+      "loss": 0.284,
+      "step": 28879
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.267089254881505e-11,
+      "loss": 0.3043,
+      "step": 28880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.034728385114093e-11,
+      "loss": 0.2808,
+      "step": 28881
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8149275458068107e-11,
+      "loss": 0.3235,
+      "step": 28882
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.607686739735215e-11,
+      "loss": 0.3049,
+      "step": 28883
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4130059694528187e-11,
+      "loss": 0.3349,
+      "step": 28884
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.230885237402113e-11,
+      "loss": 0.3255,
+      "step": 28885
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.061324545914566e-11,
+      "loss": 0.2778,
+      "step": 28886
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.043238970996016e-12,
+      "loss": 0.3399,
+      "step": 28887
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.59883292955621e-12,
+      "loss": 0.3246,
+      "step": 28888
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.280027352589812e-12,
+      "loss": 0.3031,
+      "step": 28889
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.086822256750168e-12,
+      "loss": 0.311,
+      "step": 28890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0192176564701754e-12,
+      "loss": 0.3176,
+      "step": 28891
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0772135672929583e-12,
+      "loss": 0.3046,
+      "step": 28892
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2608099981002997e-12,
+      "loss": 0.2862,
+      "step": 28893
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.570006961104653e-12,
+      "loss": 0.3138,
+      "step": 28894
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0048044651878031e-12,
+      "loss": 0.2969,
+      "step": 28895
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.652025159008645e-13,
+      "loss": 0.2879,
+      "step": 28896
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5120111879495257e-13,
+      "loss": 0.3146,
+      "step": 28897
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.280028053140542e-14,
+      "loss": 0.2844,
+      "step": 28898
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.2535,
+      "step": 28899
+    },
+    {
+      "epoch": 3.0,
+      "step": 28899,
+      "total_flos": 5634906035183616.0,
+      "train_loss": 0.4499533398256009,
+      "train_runtime": 126412.5439,
+      "train_samples_per_second": 14.63,
+      "train_steps_per_second": 0.229
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 28899,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "total_flos": 5634906035183616.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}